U.S. Pat. No. 10,286,323

DETAILED DESCRIPTION OF EMBODIMENTS

Introduction

It is generally desirable for a video game to appeal to a large number of users. However, different users have different levels of skill and/or abilities when it comes to playing video games or video games of a particular genre or type. Further, different users have different desires with respect to how challenging a video game is to play. For example, some users prefer video games that are relatively challenging. These types of users may tend to be more engaged by a video game that may require a lot of practice to master and typically may not mind repeating the same portion of the video game numerous times before being successful. In contrast, some users prefer video games that are relatively easy. These types of users may tend to be more engaged by a video game where obstacles are easily overcome and the users rarely are required to repeat a portion of the video game to be successful.

One solution to the above challenges is for video game developers to incorporate multiple static difficulty levels within a particular video game. However, there are generally a limited number of difficulty levels that a developer can add due, for example, to storage constraints, development time constraints, and the challenge of predicting a large number of difficulty levels for a large number of user preferences. Further, these difficulty levels are generally coarse because, for example, the difficulty levels are typically created by adjusting a defined set of adjustable elements (which may sometimes be referred to as “knobs” herein). Thus, because a particular user may find a particular aspect of a video game challenging, but another aspect of the video game not challenging, selecting a static difficulty level may result in an inconsistent challenge throughout the video game for the particular user. Moreover, this problem of static difficulty levels is exacerbated by the fact that another user may find different aspects of the video game challenging or easy.

Another solution that may be used in some types of competitive video games, such as racing games, is to vary the ability of the user or the user's competitor based on the relationship between the user and the user's competitor. For example, supposing that the video game is a racing game, the user's car may be made faster when the user is doing poorly and may be made slower when the user is doing well. This solution may result in what is sometimes referred to as a “rubber band effect.” This solution is often noticeable by the user because the user's vehicle will behave inconsistently based on the location of the vehicle with respect to the user's competitor. As a result, instead of appealing to a user, the user may be driven away. Further, this solution can be challenging to implement with some types of video games that do not measure a user against a specific competitor.

Embodiments presented herein include a system and method for performing dynamic difficulty adjustment. Further, embodiments disclosed herein perform dynamic difficulty adjustment using processes that are not detectable or are more difficult to detect by users compared to static and/or existing difficulty adjustment processes. In some embodiments, historical user information is fed into a machine learning system to generate a prediction model that predicts an expected duration of game play, such as for example, an expected churn rate, a retention rate, the length of time a user is expected to play the game, or an indication of the user's expected game play time relative to a historical set of users who have previously played the game. Before or during game play, the prediction model is applied to information about the user to predict the user's expected duration of game play. Based on the expected duration, the system may then utilize a mapping data repository to determine how to dynamically adjust the difficulty of the game, such as, for example, changing the values of one or more knobs to make portions of the game less difficult.

In certain embodiments, systems disclosed herein monitor user activity with respect to one or more video games to determine a user's preferences regarding game difficulty and the user's skill level with respect to playing the video games. This information may be determined based at least in part on factors that are associated with a user's engagement level. For example, a user who plays a video game for an above average length of time and who spends money while playing the video game may have a higher level of engagement than a user who plays a video game for a short period of time. As another example, a user who plays a video game for a short period of time, but who plays an above average number of play sessions may be associated with a high level of engagement, but may be classified differently than the user of the previous example.

Further, in certain embodiments described herein, users may be grouped with other users who have similar preferences into clusters. The users may be grouped based on user behavior with respect to challenges or obstacles presented in the video game. Each of the groups or clusters of users may be associated with difficulty preferences or settings for one or more video games. Using this information, one or more aspects of the video game can be dynamically adjusted to present a user of the video game with a particular difficulty level that is most likely to engage the user, or more likely to engage the user than a static set of difficulty levels. As noted above and further herein, additional or alternative embodiments described herein may determine one or more seeds or knob values for adjusting the difficulty of the video game by using one or more parameter functions or prediction models. In some cases, the prediction models may be combined with clustering. In other embodiments, the prediction models may be used instead of clustering. It should be understood that clustering is one method that may be used with embodiments of the present disclosure. However, the present disclosure is not limited to the use of clustering and certain embodiments presented herein may omit the user of clustering. For example, certain embodiments disclosed herein may use a regression model to fit historical user data without the use of clustering. After obtaining an initial version of the regression model, it can be applied to additional players to facilitate the dynamic difficulty analysis and/or adjustment.

Moreover, in certain embodiments described herein, the user's activity with respect to the video game can be monitored or reviewed to determine the user's behavior with respect to the video game. This monitoring may occur substantially in real-time, or at some period of time after the user has completed a play session. The play session may be a period of time when the user plays the video game and/or a particular attempt to play the game that ends with the user completing or failing to complete the video game or a portion thereof. For example, one play session may begin with the user initiating a new instance of game play and end with the user running out of lives in the game. As another example, one play session may begin with the user initiating the video game and ending when the user exist the video game.

In some cases, monitoring the user's behavior with respect to the video game may enable a determination of the user's skill level and desired level of challenge. Based at least in part on this information, the difficulty of the video game, or portions of the video game, can be adjusted from the initial difficulty level determined based on the associated user cluster for the user.

Advantageously, in certain embodiments, by grouping users with similar characteristics with respect to playing video games and by adjusting difficulty levels based on individual user actions with respect to the video games, a more fine-grained management of difficulty level is possible compared to systems that do not monitor user behavior to determine a difficulty level. Further, although this disclosure focuses on adjusting settings of a video game that modify the difficulty level or challenge presented by the video game, this disclosure is not limited as such. Embodiments of the present disclosure can be used to modify various aspects of game state of a video game, which may or may not affect the difficulty level of the video game. For example, in a game where weapons are randomly dropped, if it is determined that a user prefers to play a game using a particular in-game weapon, the game may be adjusted to present the preferred weapon to the user more frequently. In some cases, such as when all weapons are evenly balanced, the type of weapon dropped may not impact the difficulty of the video game and thus, such an adjustment may be based on user play styles or preferences rather than difficulty level preferences. Some other non-limiting examples of features of the video game that can be modified, which may or may not be detectable by the user can include providing extra speed to an in-game character, improving throwing accuracy of an in-game character, improving the distance or height that the in-game character can jump, adjusting the responsiveness of controls, and the like. In some cases, the adjustments may additionally or alternatively include reducing the ability of an in-game character rather than improving the ability of the in-game character. For example, the in-game character may be made faster, but have less shooting accuracy.

Further, embodiments of the systems presented herein can adjust the difficulty level of the video game substantially in real time based at least in part on the user's skill level and whether the user is successfully completing challenges within the video game. However, the present disclosure is not limited as such. For example, the difficulty level of the video game may be adjusted based at least in part on user preferences, which may or may not correspond to a user's ability. For example, some users may prefer to play video games at the most difficult settings regardless of whether they are successful at completing the video game or objectives therein. By monitoring user actions with respect to playing a video game, the difficulty level of the video game can be adjusted to match a particular user's preferences and/or skill level.

To simplify discussion, the present disclosure is primarily described with respect to a video game. However, the present disclosure is not limited as such may be applied to other types of applications. For example, embodiments disclosed herein may be applied to educational applications or other applications that may be modified based on a history of user interactivity with the application. Further, the present disclosure is not limited with respect to the type of video game. The use of the term “video game” herein includes all types of games, including, but not limited to web-based games, console games, personal computer (PC) games, computer games, games for mobile devices (for example, smartphones, portable consoles, gaming machines, or wearable devices, such as virtual reality glasses, augmented reality glasses, or smart watches), or virtual reality games, as well as other types of games.

Example Networked Computing Environment

FIG. 1Aillustrates an embodiment of a networked computing environment100that can implement one or more embodiments of a dynamic difficulty adjustment system. The networked computing environment100includes a user computing system110that can communicate with an interactive computing system130via a network104. Further, the networked computing environment100may include a number of additional user computing systems102. At least some of the user computing systems102may be configured the same as or similarly to the user computing system110.

User computing system110may include or host a video game112. In some cases, the video game112may execute entirely on the user computing system110. In other cases, the video game112may execute at least partially on the user computing system110and at least partially on the interactive computing system130. In some cases, the video game112may execute entirely on the interactive computing system130, but a user may interact with the video game112via the user computing system110. For example, the game may be a massively multiplayer online role-playing game (MMORPG) that includes a client portion executed by the user computing system110and a server portion executed by one or more application host systems (not shown) that may be included as part of the interactive computing system130. As another example, the video game112may be an adventure game played on the user computing system110without interacting with the interactive computing system130.

The video game112may include a number of knobs114that modify or affect the state of the video game112. Typically, the knobs114are variables that affect the execution or operation of the video game112. In some cases, the knobs114are state variables that directly modify the execution of the video game112. In other cases, the knobs114are seeds or seed variables that may alter the probability of an occurrence within the video game112or a random (or pseudorandom) configuration or event within the video game112. For example, one seed may correspond to and impact the generation of a level layout in the video game112. As another example, one seed may correspond to and impact the number of occurrences of item drops or the types of items dropped as a user plays the video game112. In some cases, the seed value is a value that initializes or influences a random or pseudorandom number generator. In some such cases, the random number generated based on the seed value may be utilized by one or more functions of the video game112to influence the operation of the video game112. In some cases, the seed variables may be referred to as levers, and the seed variables are non-limiting examples of various types of knobs. It should be understood that the knobs114, sometimes referred to as levers, are not limited to seeds, but can include any type of variable that may modify execution of the video game112. The system may modify execution of the video game by adjusting any type of knob or lever that can change the video game112and may conduct a knob-level analysis of the churn rate. The system is not limited to adjusting the knob or lever based on a seed value.

Generally, the knobs114are variables that relate to a difficulty level of the video game112. It should be understood that the knobs114typically include a subset of variables that modify the operation of video game112and that the video game112may include other variables not involved in the setting of the difficulty level of video game112and/or not available for modification. Further, the knobs114may include variables that modify the video game112in a manner that is not perceivable by a user or is difficult to perceive by the user. In some cases, whether or not the modification to the video game112is perceivable by the user may depend on the specific video game. For example, suppose that one knob114relates to the amount of life that an enemy in the video game112has. In some cases, modifying the value assigned to the knob114may be detectable by a user because, for example, the health of the enemy is numerically presented to the user. In such cases, the health of the enemy may remain unmodified in the difficulty level of video game112, but the difficulty level of the video game112may be modified via a different knob114. However, in some cases, modifying the health of the enemy may not be detectable by the user because, for example, the health of the enemy is not presented to the user and the amount of hits required to defeat the enemy varies within a range making it difficult for the user to determine that the health of the enemy may have been modified from one encounter with the enemy compared to another encounter with the enemy (for example, from one play through of the video game112compared to another play through of the video game112).

In some cases, the video game112may include a user interaction history repository116. The user interaction history repository116may store data or information relating to the user's interaction with the video game112. This user interaction information or data may include any type of information that can be used to determine a user's level of engagement with the video game112and/or how difficult the video game112is for the user. For example, some non-limiting examples of the user interaction information may include information relating to actions taken by the user within the video game112; a measure of the user's progress within the video game112; whether the user was successful at performing specific actions within the video game112or completing particular objectives within the video game112; how long it took the user to complete the particular objectives; how many attempts it took the user to complete the particular objectives; how much money the user spent with respect to the video game112, which may include one or both of the amount of money spent to obtain access to the video game112and the amount of money spent with respect to the video game112exclusive of money spent to obtain access to the video game112; how frequently the user accesses the video game112; how long the user plays the video game112; and the like. The user computing system110may share the user interaction information with the interactive computing system130via the network104. In some embodiments, some or all of the user interaction information is not stored by the video game112, but is instead provided to or determined by another portion of the user computing system110external to the video game112and/or by the interactive computing system130. Thus, in some embodiments, the user interaction history repository116may be optional or omitted.

The user computing system110may include hardware and software components for establishing communications over a communication network104. For example, the user computing system110may be equipped with networking equipment and network software applications (for example, a web browser) that facilitate communications via a network (for example, the Internet) or an intranet. The user computing system110may have varied local computing resources, such as central processing units and architectures, memory, mass storage, graphics processing units, communication network availability and bandwidth, and so forth. Further, the user computing system110may include any type of computing system. For example, the user computing system110may include any type of computing device(s), such as desktops, laptops, video game platforms, television set-top boxes, televisions (for example, Internet TVs), network-enabled kiosks, car-console devices, computerized appliances, wearable devices (for example, smart watches and glasses with computing functionality), and wireless mobile devices (for example, smart phones, PDAs, tablets, or the like), to name a few. In some embodiments, the user computing system110may include one or more of the embodiments described below with respect toFIG. 8andFIG. 9.

As previously discussed, it may be desirable to maintain or increase a user's level of engagement with the video game112. One solution for maintaining or increasing the user's level of engagement with the video game112includes setting or adjusting a difficulty level of the video game112based at least in part on a user's skill and a user's desired level of challenge when playing the video game112. The interactive computing system130can determine a level of difficulty for the video game112for a particular user and can modify the difficulty of the video game112based on the determination. This determination, as will be described in more detail below, of the difficulty level may be made based at least in part on user interaction information with respect to the video game112and/or other video games accessible by the user.

Interactive computing system130may include a number of systems or subsystems for facilitating the determination of the difficulty level for the video game112for a particular user and the modification of the difficulty level based on the determination. These systems or subsystems can include a difficulty configuration system132, a user clustering system134, a seed evaluation system136, a user data repository138, a retention analysis system140, a model generation system146, and a mapping data repository144. Each of these systems may be implemented in hardware, and software, or a combination of hardware and software. Further, each of these systems may be implemented in a single computing system comprising computer hardware or in one or more separate or distributed computing systems. Moreover, while these systems are shown inFIG. 1Ato be stored or executed on the interactive computing system130, it is recognized that in some embodiments, part or all of these systems can be stored and executed on the user computing system110.

In some embodiments, when the user computing system110is connected or in communication with the interactive computing system130via the network104, the interactive computing system130may perform the processes described herein. However, in some cases where the user computing system110and the interactive computing system130are not in communication, the user computing system110may perform certain processes described herein using recent game play of the user that may be stored in the user interaction history repository116.

The difficulty configuration system132sets or adjusts the difficulty level of a video game112. In some cases, the difficulty configuration system132may set or adjust the difficulty level of the video game112by providing or adjusting values for one or more of the knobs114, which are then fed into or provided to the video game112. In some cases, the difficulty configuration system132sets or adjusts every available knob114each time a setting or an adjustment of a difficulty level is made. In other cases, the difficulty configuration system132may set or adjust less than every available knob114when setting or adjusting a difficulty level of the video game112.

In some cases, the difficulty configuration system130may modify the difficulty of a portion of the video game112without constraint. However, in some other cases, the difficulty configuration system132may modify the difficulty of a portion of the video came112within a set of constraints that may be specified by a developer, a set of rules, or that are relative to other portions of the video game112. For example, in some cases, the difficulty configuration system132may reduce the difficulty level of a particular portion of the video game112to no more than a difficulty threshold specified by a proceeding portion of the video game112. Thus, in some cases, a reduction in the difficulty level of the particular portion of the video game112will not result in the particular portion of the video game112becoming easier than the proceeding portion112. The difficulty adjustment may be turned off completely in some situations, such as during a tournament.

The model generation system146can use one or more machine learning algorithms to generate one or more prediction models or parameter functions. One or more of these prediction models may be used to determine an expected value or occurrence based on a set of inputs. For example, a prediction model can be used to determine an expected churn rate or a probability that a user will cease playing the video game112based on one or more inputs to the prediction model, such as, for example, historical user interaction information for a user. As another example, a prediction model can be used to determine an expected amount of money spent by the user on purchasing in-game items for the video game based on one or more inputs to the prediction model. In some cases, the prediction model may be termed a prediction model because, for example, the output may be or may be related to a prediction of an action or event, such as the prediction the user continues to play the video game112. A number of different types of algorithms may be used by the model generation system146. For example, certain embodiments herein may use a logistical regression algorithm. However, other algorithms are possible, such as a linear regression algorithm, a discrete choice algorithm, or a generalized linear algorithm.

The machine learning algorithms can be configured to adaptively develop and update the models over time based on new input received by the model generation system146. For example, the models can be regenerated on a periodic basis as new user information is available to help keep the predictions in the model more accurate as the user information evolves over time. The model generation system146is described in more detail herein. After a model is generated, it can be provided to the retention analysis system140.

Some non-limiting examples of machine learning algorithms that can be used to generate and update the parameter functions or prediction models can include supervised and non-supervised machine learning algorithms, including regression algorithms (such as, for example, Ordinary Least Squares Regression), instance-based algorithms (such as, for example, Learning Vector Quantization), decision tree algorithms (such as, for example, classification and regression trees), Bayesian algorithms (such as, for example, Naive Bayes), clustering algorithms (such as, for example, k-means clustering), association rule learning algorithms (such as, for example, Apriori algorithms), artificial neural network algorithms (such as, for example, Perceptron), deep learning algorithms (such as, for example, Deep Boltzmann Machine), dimensionality reduction algorithms (such as, for example, Principal Component Analysis), ensemble algorithms (such as, for example, Stacked Generalization), and/or other machine learning algorithms.

The retention analysis system140can include one or more systems for determining a predicted churn or retention rate for a user based on the application of user interaction data for the user to a prediction model generated by the model generation system140. In some cases, the difficulty configuration system132may use the predicted retention rate determined by the retention analysis system140to determine an adjustment to the difficulty of the video game112. In some embodiments the adjustments to the difficulty are determined using data in a mapping data repository144to determine which features of the game to change and how to change the features.

The mapping data repository144can include one or more mappings between the output of a prediction model and a difficulty level of the video game112, which may be used by, for example, the difficulty configuration system132to determine how to modify the video game112to adjust the difficulty of the video game112. For example, if the user's predicted level of churn is “high,” then the mapping data repository144may include a mapping between “high” and data to adjust the game to be “easy” such that one or more knobs or seeds associated with the game are set to values that make the game easier to play. As another example, if the user's predicted level of churn is 65%, then the mapping data repository144may include a mapping between “60-70%” to adjust a specific subset of the one or more knobs or seeds associated with the game to sets of values that make the game less difficult to play, but not the easiest to play. Alternatively, or in addition, the mapping may be between the output of the parameter function and one or more values for one or more knobs or seeds that can be used to modify the difficulty of the video game112.

Further, generation and application of the parameter functions and their use in adjusting the difficulty level of the video game112will be described in further detail below with respect to the retention analysis system140. In certain embodiments, the difficulty configuration system132may be or may include the model generation system146. Moreover, in some cases, the difficulty configuration system132may be or may include the retention analysis system140. As stated above, one non-limiting example of a machine learning algorithm that can be used herein is a clustering algorithm. The user clustering system134may facilitate execution of the clustering algorithm. The user clustering system134groups or divides a set of users into groups based at least in part on each user's skill level with respect to the video game112or other video games accessed by the users. Alternatively, or in addition, the user clustering system134may group or cluster the users based on one or more criteria associated with one or more of the users that impacts the users' engagement level with the video game112or other video games accessed by the users. Furthermore, the user clustering system134may identify or determine a set of difficulty preferences to associate with each user cluster identified or generated by the user clustering system134.

The seed evaluation system136, which may also be referred to as a lever evaluation system, evaluates a difficulty or a challenge provided by the video game112in response to a seed value. For example, the seed evaluation system136can determine how challenging a particular level or portion of the video game (such as a dungeon) generated in response to a particular seed value in a video game112is based on how well a group of users do playing the video game112when the particular seed value is utilized. Advantageously, in certain embodiments, by evaluating the challenge provided by a particular seed value, the difficulty level of the video game112can be refined by adding or removing the seed value to a set of available seed values for a particular difficulty level. For example, if it is determined that a seed value causes users to fail at an 80% rate, the seed value may be associated with the harder difficulty level than another seed value that causes users to fail at a 20% rate.

The user data repository138can store user interaction information associated with one or more users' interaction with the video game112and/or one or more other video games. This user interaction information can be obtained over one or more play sessions of the video game112. Further, the user data repository138can store user cluster information associated with one or more user clusters generated by the user clustering system134. In some cases, at least some of the data stored in the user data repository128may be stored at a repository of the user computing system110. Each of the repositories described herein may include non-volatile memory or a combination of volatile and nonvolatile memory.

The network104can include any type of communication network. For example, the network104can include one or more of a wide area network (WAN), a local area network (LAN), a cellular network, an ad hoc network, a satellite network, a wired network, a wireless network, and so forth. Further, in some cases, the network104can include the Internet.

Example Model Generation System

FIG. 1Billustrates an embodiment of the model generation system146ofFIG. 1A. The model generation system146may be used to determine one or more prediction models160based on historical data152for a number of users. Typically, although not necessarily, the historical data152includes data associated with a large number of users, such as hundreds, thousands, hundreds of thousands, or more users. However, the present disclosure is not limited as such, and the number of users may include any number of users. Further, the historical data152can include data received from one or more data sources, such as, for example, an application host system (not shown) and/or one or more user computing systems102. Further, the historical data152can include data from different data sources, different data types, and any data generated by one or more user's interaction with the video game112. In some embodiments, the historical data152may include a very large number of data points, such as millions of data points, which may be aggregated into one or more data sets. In some cases, the historical data152may be accessed from a user data repository138. In some embodiments, the historical data152is limited to historical information about the video game, but in other embodiments, the historical data152may include information from one or more other video games. Further, in some embodiments, one or more subsets of the historical data a limited by a date restriction, such as for example, limited to include only data from the last 6 months.

The model generation system146may, in some cases, also receive feedback data154. This data may be received as part of a supervised model generation process that enables a user, such as an administrator, to provide additional input to the model generation system146that may be used to facilitate generation of the prediction model160. For example, if an anomaly exists in the historical data152, the user may tag the anomalous data enabling the model generation system146to handle the tagged data differently, such as applying a different weight to the data or excluding the data from the model generation process.

Further, the model generation system146may receive control data156. This control data156may identify one or more features or characteristics for which the model generation system146is to determine a model. Further, in some cases, the control data156may indicate a value for the one or more features identified in the control data156. For example, suppose the control data156indicates that a prediction model is to be generated using the historical data152to determine a length of time that the users played the video game112. If the amount of time each user played the game is known, this data may be provided as part of the control data156, or as part of the historical data152. As another example, if the prediction model is to be generated to estimate a retention rate as determined, for example, based on whether the users played the video game112for a threshold period of time or continue to play the video game112after a particular threshold period of time, the control data156may include the retention rate for the users whose data is included in the historical data152.

The model generation system146may generally include a model generation rule set170for generation of the prediction model160. The rule set170may include one or more parameters162. Each set of parameters162may be combined using one or more mathematical functions to obtain a parameter function. Further, one or more specific parameters may be weighted by the weights164. In some cases, the parameter function may be obtained by combining a set of parameters with a respective set of weights164. The prediction model160and/or the respective parameters162of the prediction models160may be derived during a training process based on particular input data, such as the historical data152, feedback data154, and control data156, and defined output criteria, which may be included with the control data156, used for training purposes. The model generation rule set170can define the specific machine learning rules and/or algorithms the model generation system146uses to generate the model based on a defined objective function, such as determining a churn rate. In some embodiments, initial parameters162and weights164can be manually provided during the initiation of the model generation process. The parameters162and weights164can be updated and modified during the model generation phase to generate the prediction model160.

The model generation system146can filter and categorize the historical data sets according to various characteristics and parameters of the data. For example, the data can be categorized by the data source (such as, for example, game application data, host application data, or user profile data), information type (such as, for example, gameplay information, transaction information, interaction information, game account information), or other categories associated with the data. The model generation system146can filter the information to identify the information for further processing. In some embodiments, the model generation system146is configured to filter and separate the historical data152into a plurality of data types or categories before further processing. Moreover, in some cases, some of the historical data152may be filtered out or removed from the historical data152based on the data being associated with a relevance that does not satisfy a threshold relevance as determined by the model generation system146.

Optionally, one or more of the prediction models160may be associated with a penalty166. These penalties166may be used to facilitate the generation of or selection of a particular prediction model160based on one or more factors that are used to derive the penalty. For example, the mathematical complexity or the number of parameters included in a particular prediction model160may be used to generate a penalty for the particular prediction model160, which may impact the generation of the model and/or a selection algorithm or a selection probability that the particular prediction model160is selected.

After the prediction model160has been generated, the model can be used during runtime of the retention analysis system140and/or the difficulty configuration system132to adjust the difficulty of the video game112. In some cases, the adjustment of the difficulty may be dynamic and may occur during a user's interaction with the video game112. Further, in some cases, the difficulty adjustment may occur in real-time or near real-time.

Example Retention Analysis System

FIG. 1Cillustrates an embodiment of a retention analysis system140ofFIG. 1A. The retention analysis system140can apply or use one or more of the prediction models160generated by the model generation system146. Although illustrated as a separate system, in some cases, the retention analysis system140may be included as part of the difficulty configuration system132. The retention analysis system140may use one or more prediction models160A,160B,160N (which may be referred to collectively as “prediction models160” or in the singular as “prediction model160”) to process the input data172to obtain the output data174.

The retention analysis system140may apply the prediction model(s)160during game play. In some embodiments, the prediction models160are applied at the beginning of the game to determine how to adjust the difficulty of the entire game. In other embodiments, the prediction models160are applied at different times during the game and/or at different stages in the game. During determination of a difficulty level for one or more portions of the video game112, the retention analysis system140receives input data172that can be applied to one or more of the prediction models160. The input data172can include one or more pieces of data associated with a user who is playing the video game112. This data may include user interaction data for the video game112, profile data for the user, and any other data that may be applied to the prediction model160to determine a retention or churn rate for the user. In some embodiments, the input data172can be filtered before it is provided to the retention analysis system140.

In some embodiments, a single prediction model160may exist for the retention analysis system140. However, as illustrated, it is possible for the retention analysis system140to include multiple prediction models160. The retention analysis system140can determine which detection model, such as any of models160A-N, to use based on input data172and/or additional identifiers associated with the input data172. Additionally, the prediction model160selected may be selected based on the specific data provided as input data172. The availability of particular types of data as part of the input data172can affect the selection of the prediction model160. For example, the inclusion of demographic data (for example, age, gender, first language) as part of the input data may result in the use of prediction model160A. However, if demographic data is not available for a particular user, then prediction model1608may be used instead.

As mentioned above, one or more of the prediction models160may have been generated with or may be associated with a penalty166. The penalty may be used to impact the generation of the model or the selection of a prediction model for use by the retention analysis system140.

The output data174can be a retention rate or churn rate associated with a prediction that a user ceases to play the video game112. For example, in some embodiments, the retention rate may be between 0 and 100 indicating the predicted percentage of users associated with similar or the same data as included as input data172who would cease to play the video game112within a threshold time period. In some cases, the output data174may also identify a reason for the retention rate. For example, the retention analysis system140may indicate that the 90% retention rate for a particular user is based at least in part on the amount of money spent while playing the video game112. However, the retention analysis system140may indicate that the 90% retention rate for another user may be based at least in part on the below freezing temperature in the geographic region where the user is located. As another example, the retention analysis system140may indicate that the 20% retention rate for a user may be based at least in part on the below 25% win ratio.

The prediction models160A,160B,160N may generally include a set of one or more parameters162A,162B,162N, respectively (which may be referred to collectively as “parameters162”). Each set of parameters162(such as parameters162A) may be combined using one or more mathematical functions to obtain a parameter function. Further, one or more specific parameters from the parameters162A,162B,162N may be weighted by the weights164A,164B,164N (which may be referred to collectively as “weights164”). In some cases, the parameter function may be obtained by combining a set of parameters (such as the parameters162A) with a respective set of weights164(such as the weights164A). Optionally, one or more of the prediction models160A,160B,160N may be associated with a penalty166A,1668,166N, respectively (which may be referred to collectively as “penalties166”).

Example Machine Learning Process

FIG. 2presents a flowchart of an embodiment of a machine learning process200. The process200can be implemented by any system that can generate one or more parameter functions or prediction models that include one or more parameters. In some cases, the process200serves as a training process for developing one or more parameter functions or prediction models based on historical data or other known data. The process200, in whole or in part, can be implemented by, for example, an interactive computing system130, a difficulty configuration system132, a user clustering system134, a retention analysis system140, a model generation system146, or a user computing system110, among others. Although any number of systems, in whole or in part, can implement the process200, to simplify discussion, the process200will be described with respect to particular systems. Further, it should be understood that the process200may be updated or performed repeatedly over time. For example, the process200may be repeated once per month, with the addition or release of a new video game, or with the addition of a threshold number of new users available for analysis or playing a video game112. However, the process200may be performed more or less frequently.

The process200begins at block202where the model generation system146receives historical data152comprising user interaction data for a number of users of the video game112. This historical data152may serve as training data for the model generation system146and may include user demographics or characteristics, such as age, geographic location, gender, or socioeconomic class. Alternatively, or in addition, the historical data152may include information relating to a play style of one or more users; the amount of money spent playing the video game112; user success or failure information with respect to the video game112(for example, a user win ratio); a play frequency of playing the video game112; a frequency of using particular optional game elements (for example, available boosts, level skips, in-game hints, power ups, and the like); the amount of real money (for example, U.S. dollars or European euros) spent purchasing in-game items for the video game112; and the like. Further, in some cases, the historical data152may include data related to the video game112, such as one or more seed values used by users who played the video game112. Additional examples of data related to the video game112that may be received as part of the historical data152may include settings for one or more knobs or state variables of the video game112, the identity of one or more difficulty levels for the video game112used by the users, the type of the video game112, and the like.

At block204, the model generation system146receives control data156indicating a desired prediction for the number of users corresponding to the historical data. This control data156may indicate one or more features or characteristics for which the model generation system146is to determine a model. Alternatively, or in addition, the control data156may include a value for the features or characteristics that are associated with the received historical data152. For example, the control data156may identify churn rate, or retention rate, as the desired feature to be predicted by the model that is to be generated by the model generation system146. The churn rate or retention rate may correspond to a percentage of users associated with the historical data152that ceased playing the video game112. Further, the control data156may identify a retention rate associated with the historical data. For example, the control data156may indicate that the retention rate is 60% for certain of the users whose data is included in the historical data152. In some embodiments, the control data156may include multiple characteristics or features to be predicted by the model to be generated by the model generation system146. For example, the control data156may identify both a retention rate and a reason for the retention rate (such as the difficulty of the video game112being too low or too high for the users whose data was provided as part of the historical data152at block202), or a retention rate and an average monetary amount spent by the users whose data was provided as the historical data152.

At block206, the model generation system146generates one or more prediction models160based on the historical data152and the control data156. The prediction models160may include one or more variables or parameters162that can be combined using a mathematical algorithm or model generation ruleset170to generate a prediction model160based on the historical data152and, in some cases, the control data156. Further, in certain embodiments, the block206may include applying one or more feedback data154. For example, if the prediction model160is generated as part of a supervised machine learning process, a user (for example, an administrator) may provide one or more inputs to the model generation system146as the prediction model160is being generated and/or to refine the prediction model160generation process. For example, the user may be aware that a particular region or geographic area had a power outage. In such a case, the user may supply feedback data154to reduce the weight of a portion of the historical data152that may correspond to users from the affected geographic region during the power outage. Further, in some cases, one or more of the variables or parameters may be weighted using, for example, weights164. The value of the weight for a variable may be based at least in part on the impact the variable has in generating the prediction model160that satisfies, or satisfies within a threshold discrepancy, the control data156and/or the historical data152. In some cases, the combination of the variables and weights may be used to generate a prediction model160.

Optionally, at block208, the model generation system146applies a penalty to or associates a penalty166with at least some of the one or more prediction models160generated at block206. The penalty associated with each of the one or more prediction models160may differ. Further, the penalty for each of the prediction models160may be based at least in part on the model type of the prediction model160and/or the mathematical algorithm used to combine the parameters162of the prediction model160, and the number of parameters included in the parameter function. For example, when generating a prediction model160, a penalty may be applied that disfavors a very large number of variables or a greater amount of processing power to apply the model. As another example, a prediction model160that uses more parameters or variables than another prediction model may be associated with a greater penalty166than the prediction model that uses fewer variables. As a further example, a prediction model that uses a model type or a mathematical algorithm that requires a greater amount of processing power to calculate than another prediction model may be associated with a greater penalty than the prediction model that uses a model type or a mathematical algorithm that requires a lower amount of processing power to calculate.

The model generation system146, at block210, based at least in part on an accuracy of the prediction model160and any associated penalty, and selects a prediction model160. In some embodiments, the model generation system146selects a prediction model160associated with a lower penalty compared to another prediction model160. However, in some embodiments, the model generation system146may select a prediction model associated with a higher penalty if, for example, the output of the prediction model160is a threshold degree more accurate than the prediction model associated with the lower penalty. In certain embodiments, the block210may be optional or omitted. For example, in some cases, the prediction models160may not be associated with a penalty. In some such cases, a prediction model may be selected from a plurality of prediction models based on the accuracy of the output generated by the prediction model.

Example Difficulty Based Seed Selection Process

FIG. 3presents a flowchart of an embodiment of a difficulty based seed selection process300. The process300can be implemented by any system that can select a seed value for adjusting a difficulty of a video game based at least in part on the output of a prediction function or a parameter function. The process300, in whole or in part, can be implemented by, for example, an interactive computing system130, a difficulty configuration system132, a user clustering system134, a model generation system146, a retention analysis system140, or a user computing system110, among others. Although any number of systems, in whole or in part, can implement the process300, to simplify discussion, the process300will be described with respect to particular systems. Further, it should be understood that the process300may be updated or performed repeatedly over time. For example, the process300may be repeated for each play session of a video game112, for each round of the video game112, each week, each month, for every threshold number of play sessions, for every threshold number of times a user loses or fails to complete an objective, each time a win ratio drops below a threshold level, and the like. However, the process300may be performed more or less frequently.

The process300begins at block302where the retention analysis system140receives a set of input data (such as the input data172) comprising user interaction data for a user of the video game112. This input data172is typically, although not necessarily, user specific data. Further, the set of input data172may include both historical user interaction data and recent user interaction data for the user. Historical user interaction data may include user interaction data from a noncurrent play session and/or user interaction data that satisfies a threshold age or that is older than a particular threshold time period. For example, the historical user interaction data may include user interaction data that is at least a week or a month old. Alternatively, or in addition, the historical user interaction data may include data from play sessions that are more than 5 or 10 play sessions old.

Conversely, the recent user interaction data may include user interaction data that satisfies a threshold age or that is more recent than a particular threshold time period. For example, the recent user interaction data may include user interaction data that is less than a week or a month old. Alternatively, or in addition, the recent user interaction data may include data from play sessions that are less than 3, 5, or 10 play sessions old.

In some embodiments, the historical user data and the recent user data may be weighted differently within the prediction model160. In some cases, each parameter162within the prediction model160may be repeated. For example, one version of the parameter may be based on the historical user data and may be associated with one weight164and another version of the parameter may be based on the recent user data and may be associated with a different weight164. Moreover, in some implementations, the weight may be applied on a sliding scale or a graduated basis. For example, more recent historical user data may be associated with a higher weight164and less recent historical user data.

At block304, the retention analysis system140, using a parameter function or a prediction model160, may determine a predicted churn rate for the user based at least in part on the set of input data172received at block302. The predicted churn rate for the user may be determined at the beginning of game play, and/or at various stages or at various time frames during game play. The determined or calculated output174of the prediction model160may be a predicted churn rate or an expected churn for users with the same or substantially similar input data as the set of input data received at block302. Further, the churn rate may indicate the percentage of users or the likelihood that a user will continue to play or not continue to play the video game112based on the set of input data received at block302. For example, if a churn rate of 75% is output by the retention analysis system140, then it may be estimated that there is a 75% chance that the user will not continue to play the video game112. Alternatively, or in addition, the churn rate of 75% may indicate that 75% of users with the same or similar user interaction data that is associated with the set of input data received at the block302will cease playing the video game112generally, after a certain amount of time, or at a specific point in the game. In some embodiments, the prediction model160may predict the amount of time the user is predicted to play the video game112over a particular time period or until ceasing to play the video game112. This determination may be for a particular play session or for a number of play sessions and may be instead of or in addition to determining the retention rate.

Optionally, at block306, using the prediction model160, the retention analysis system140determines a predicted reason for the predicted churn rate for the user based on the set of input data172. For example, the parameter function may determine based on an input data indicating a win ratio below a threshold that the churn rate calculated by the parameter function is based at least in part on the difficulty of the video game112.

At block308, based on the predicted churn rate for the user determined at block304, the difficulty configuration system132selects a seed associated with a particular level of difficulty for a portion of the video game. Selecting the seed may include accessing mapping data at the mapping data repository144. This mapping data may indicate a mapping between one or more churn rates and one or more difficulty levels for the video game112. In some cases, the churn rate above a particular threshold level may be mapped to one or more seed values that may reduce the difficulty of the video game112. Further, in some cases, a churn rate may be mapped to different sets of seed values associated with different difficulty levels. In such cases, the set of seed values associated with a particular difficulty level may be selected based on an additional output174of the retention analysis system140. For example, the retention analysis system140may output a churn rate and a reason for the churn rate for a particular user. Such an example, a churn rate above a particular threshold may cause a new seed value to be selected. Further, the reason for the churn rate may be used to select a plurality of seed values associated with different difficulty levels. The reason for the churn rate may be associated with particular numeric values. For example, a churn rate above a particular threshold caused by the video game112being too easy may be associated with one numeric value in a churn rate above a particular threshold caused by the video game112being too hard may be associated with another numeric value.

In some embodiments, the retention analysis system140may output the one or more factors or data from the input data172that had the most impact or a threshold amount of impact in determining the retention rate for the user. Using this information, the difficulty configuration system132may identify one or more changes to state variables of the video game112, or one or more seed values, that correspond to the factors used to generate the retention rate.

In some cases, the mapping between the output from the retention analysis system140and the seed values may be a multistage mapping. For example, a first mapping may be between the churn rate output by the retention analysis system140and a plurality of seed values, and a second mapping may be between another output of the retention analysis system140(such as a predicted reason for the churn rate) and a particular seed value or subset of seed values from the plurality of seed values. In another implementation, the multistage mapping may include a first stage mapping between the churn rate and a particular difficulty level, and a second stage mapping between the particular difficulty level and one or more seed values or knob values.

In some cases, the seeds are adjustments to knobs were state variables used to modify execution of the video game112. In other cases, the seeds are values used in random or pseudorandom number generators that are used to modify the probability or the occurrence rate of one or more events within the video game112. Further, in some cases, the seeds are values used to affect the generation of a level or a portion of the video game112and/or the location of a user playable character within the video game112.

In some cases, one or more of the above embodiments may be combined with clustering to facilitate identifying users who may prefer to play video games that are more or less difficult. Some example embodiments of clustering the may be used with the present disclosure are described in more detail below.

Example Clustering Embodiments

In some embodiments, a clustering process may be used to group users who share one or more characteristics that may be used to identify difficulty preferences for a video game112. Clusters may include one or more users that share one or more characteristics. Difficulty preferences can be associated with each cluster to facilitate adjusting the difficult of the video game112. A user whose characteristics match a particular cluster, may be assigned to the particular cluster. A difficulty level of the video game112can be adjusted for the user based on difficulty preferences associated with the cluster. Certain non-limiting example processes are described below that enable difficult adjustment using clustering.

Example Cluster Creation Process

FIG. 4presents a flowchart of an embodiment of a cluster creation process400. The process400can be implemented by any system that can create a plurality of clusters or groups of users based on each user's interaction with a video game and the level of engagement associated with each user. For example, the process400, in whole or in part, can be implemented by an interactive computing system130, a difficulty configuration system132, a user clustering system134, or a user computing system110, among others. Although any number of systems, in whole or in part, can implement the process400, to simplify discussion, the process400will be described with respect to particular systems. Further, it should be understood that the process400may be updated or performed repeatedly over time. For example, the process400may be repeated once per month, with the addition or release of a new video game, or with the addition of a threshold number of new users available for analysis or playing a video game112. However, the process400may be performed more or less frequently.

The process400begins at block402where the user clustering system134identifies a set of users of the video game112. To simplify discussion, the process400is primarily described with respect to a single video game, such as the video game112. However, this disclosure is not limited as such, and the process400can be implemented for a plurality of video games. In some cases, each of the plurality of video games may be of the same genre or may share one or more characteristics in common. In other cases, the plurality of video games may be distributed across a number of genres. The genres may be based on theme and/or game type (for example, an open world game, a role-playing game, a first-person shooter, a side scrolling game, a simulation, a space fighter, a western, and the like). Further, in some cases, the process400includes analyzing data across additional video games to confirm or refine the clusters created based on the analysis of a video game.

At block404, for each user identified at the block402, the user clustering system134monitors the user's interaction with the video game112over time to obtain user interaction data for the user. This monitoring can be done by reviewing sets of user interaction data for the user from different time periods or by pulling data from the video game112in real time and storing the data for later review. This user interaction data can include any of the information previously described with respect toFIG. 1A. Further, the user interaction data may include data relating to the users progress within the video game; the action taken by the user when the user succeeds in completing a level or objective; the action taken by the user when the user does not succeed in completing a level or objective; differences in actions taken by the user based on the length of time it takes the user to succeed at an objective; the length of time the user plays the video game each time or on average when the user plays the video game; whether the user typically plays the game for short periods of time or long periods of time; whether the user spends real world currency (as opposed to in-game currency) when playing the video game, which may be used as a factor to identify the users level of engagement (for example, a user spends money while playing a video game is more likely to play the game again compared to a user who does not spend money or playing a video game); and any other criteria that can be used to measure the users level of engagement with the video game and/or the user's action in response to the level of challenge that the video game presents to the user. Further, user interaction data may also include information relating to the type of user computing system110used by the user to access the video game; differences, if any, and how the user interacts with the video game based on the type of user computing system110used to access the video game; whether the user uses multiple user computing systems110to access the video game; and the like.

At block406, the user clustering system134filters out users whose user interaction data does not satisfy a minimum set of interaction criteria. This minimum set of interaction criteria may be related to the length of time that the user played the game or to whether the user played the video game for multiple play sessions. For example, a user who played the video game less than a threshold amount of time or for less than a threshold number of play sessions may not have provided sufficient data to determine whether the video game's difficulty impacted the user's level of engagement. Further, the minimum set of interaction criteria may be related to the type of actions the user has taken in the video game112or the progress the user has made in the video game112. In some cases, users whose user interaction data does not satisfy the minimum set of interaction criteria may be retained, but weighted lower compared to user interaction data for users that does satisfy the minimum set of interaction criteria. In some embodiments, the block406may be optional or omitted.

For each remaining user from the set of users, the user clustering system134determines a level of engagement for the user based on the user interaction data for that user at block408. Determining the level of engagement for the user may include determining whether the user continues to play or stops playing a video game based on the user's progress within the video game. Further, determining the level of engagement for the user may include determining whether and/or how much money the user spends while playing the video game. In some cases, determining the level of engagement for the user may include determining a probability that the user will play the video game again based on the user interaction data collected for the user. In some cases, determining the level of engagement for the user may include determining the user's skill with respect to the video game. Further, performing the operations associated with the block408may include applying one or more machine learning algorithms using the user interaction data as input to determine a probability that the user continues to play or stops playing the video game based on the amount of challenge presented to the user by the video game.

At block410, the user clustering system134determines a number of user clusters based on the user interaction data and the level of engagement for each user of the remaining users. Determining the user clusters may include grouping users based on their behavior as determined from the user interaction data monitored at the block404. Further, grouping the users into the user clusters may include identifying characteristics associated with each user based on the user interaction data collected for the users that indicate a level of engagement with the video game. For example, suppose that the system determines that a number of users typically play the video game for less than 30 minutes and tend to succeed at an objective in less than two attempts over the course of a threshold number of play sessions and/or objectives. Further, suppose that these users typically stop playing the video game after reaching some number of objectives that require more than two attempts to complete the objectives. This number of users may be clustered together in a user cluster for users who tend to play short game sessions and tend to prefer video games that are not challenging. In contrast, another group of players who tend to play video games for two hours at a time and continue to play the video game despite objectives taking several attempts to complete may be clustered together in a separate user cluster. One or more machine learning algorithms may be used to identify the clusters of users based at least in part on the user interaction data for the set of users.

In some cases, block410may include generating subclusters within each cluster. For example, one cluster may include users who tend to prefer video games that are particularly challenging. Within this cluster, there may be two subclusters. One subcluster may be for users who tend to prefer video games that remain challenging throughout the course of the entire game. Another subcluster may be for users who tend to prefer video games that are challenging in the beginning, but may become less challenging over time as may occur with video games that are difficult to master, but are less difficult once the user masters the skills required by the video game.

At block412, the difficulty configuration system132associates difficulty preferences with each user cluster based on the user interaction data associated with the users in the user cluster. In some embodiments, a machine learning model may be generated by, for example, the model generation system146, to determine which values for difficulty preferences are associated with longer game play compared to other values for the difficulty preferences. The difficulty preferences may identify whether users associated with the user cluster prefer to play a video game that is difficult, easy, or some gradation in between. Further, the difficulty preferences may identify one or more values resetting one or more knobs were game state variables associated with the video game.

Example Cluster Assignment Process

FIG. 5presents a flowchart of an embodiment of a cluster assignment process500for a user. The process500can be implemented by any system that can identify a user cluster with which to associate a user based on the user's interaction with a video game. For example, the process500, in whole or in part, can be implemented by an interactive computing system130, a difficulty configuration system132, a user clustering system134, or a user computing system110, among others. Although any number of systems, in whole or in part, can implement the process500, to simplify discussion, the process500will be described with respect to particular systems. Further, it should be understood that the process500may be updated or performed repeatedly over time. For example, the process500may be repeated once per month, after a threshold number of play sessions by the user since a prior performance of the process500, or after the user plays a new video game. However, the process500may be performed more or less frequently. Furthermore, the process500may be performed in real time or may be performed in advance of a particular event. For example, the process500may be performed upon user registration with a game deployment system or upon the user accessing the video game112.

The process500begins at block502where the user clustering system134identifies a user of the video game112. The user may be identified based on user account information, such as a user login, or based on information associated with an avatar of the user, such as a screen name. Alternatively, or in addition, a user may be identified based on information associated with a user computing system110of the user, such as an Internet protocol (IP) address.

At block504, the user clustering system134monitors the user's interaction with the video game112over a time period to obtain user interaction data for the user. This monitoring can be done by reviewing sets of user interaction data for the user from different time periods or by pulling data from the video game112in real time and storing the data for later review. Typically, the time period is a historical time period that may include the user's interaction with the video game112over multiple play sessions. Further, the length of the time period may be selected to satisfy or exceed a minimum time threshold. For example, the time period may be selected to be at least, or to exceed, one month, two months, half a year, and the like. In some cases, instead of or in addition to monitoring the user's interaction with the video game over a time period, the user clustering system134may be configured to monitor the user's interaction over a threshold number of play sessions. For example, the user clustering system134may be configured to monitor the first number (for example, five, ten, twelve, and the like) of play sessions of the user or the most recent number of play sessions of the user. In some cases, the block504may include monitoring the user's interaction with a plurality of video games. The plurality of video games may be video games of the same type as the video game112. In other cases plurality of video games may not be limited to a particular type of video game.

The user clustering system134accesses cluster definitions for a set of clusters at block506. Accessing the cluster definitions for the set of clusters may include accessing a user data repository138. The cluster definitions may include a set of characteristics that correlate to or are derived from user interaction data for a set of users.

Using the cluster definitions accessed at the block506and the user interaction data obtained at the block504, the user clustering system134identifies a cluster from the set of clusters at the block508. Identifying the cluster from the set of clusters may include matching characteristics of the user interaction data with characteristics associated with each of the set of clusters. For example, if the user interaction data indicates that the user plays the video game112for several hours at a time when successfully completing objectives and ceases to play the video game112within about 10 minutes on average when failing to complete an objective, then the user clustering system134may determine that the user is a player who spends significant time playing video games and does not enjoy significant challenge. Continuing this example, the user clustering system134may identify the user cluster from the set of clusters that is associated with players who play video games for a significant amount of time and tend to enjoy games with less than a threshold level of difficulty.

In some cases, the determination of the cluster from the set of clusters may include identifying how difficult the user finds the video game112based on the user interaction data obtained at the block504. Further, the determination of the cluster of a set of clusters may include identifying the user's actions or reactions to events within the video game including events relating to the success or failure of the user in overcoming challenges within the video game112. Engagement characteristics or user interaction characteristics for the user with respect to the video game may be determined based on the user's behavior with respect to the success or failure of satisfying objectives within the video game112. These engagement characteristics and/or other characteristics related to the user that are derived from the user's interaction with the video game may be compared to characteristics associated with the set of clusters to identify a corresponding cluster to associate with the user.

In some embodiments, the engagement characteristics may be presented to the user and, in response, the user clustering system134may receive input from the user regarding the engagement characteristics. For example, the user may indicate whether the user agrees with the analysis. As another example, the user may indicate that he or she was experimenting with a new play style that the user does or does not plan to continue using. The user clustering system134may use the user input to adjust or confirm its selection of a particular user cluster. In some cases, the user input may be weighted based on the amount of data the user clustering system134has obtained at block504. For example, the user input may be weighted more heavily for users with a little history (such as two or three play sessions) and weighted less heavily for users with a significant amount of history (such as fifty or a hundred play sessions).

In some cases, the interactive computing system130may cause sliders, or some other user interface element, to be displayed to the user to indicate the user's engagement characteristics on a spectrum. For example, the slider may indicate that the user tends to attempt challenges more often than average or that the user on average tends to be more successful at certain challenges than other users. Although the analysis of the user interaction data may be presented to the user, the user may not be informed that the information is being used to adjust the difficulty level of at least portions of the video game112.

After the user has played a portion of the video game112, the user clustering system134may question the user to help determine user preferences or to obtain information regarding how the user viewed the difficulty of the portion of the video game112. The user may be questioned after the user indicates that the user is ending a play session. Thus, the user can be questioned regarding his or her experience without interrupting the user's play experience. Further, the user clustering system134may parse chat message data of the user to determine the user's engagement level and/or how difficult the user finds the video game112.

At block510, the user clustering system134associates the user with the identified cluster. Associating the user with the identified cluster may include storing an association between the user and the identified cluster at the user data repository138.

In some embodiments, the process500may be used to determine how likely a user is to stop playing the video game112. This determination can sometimes be referred to as a “churn rate” or “churn” and can be associated with how often a user switches video games or stops playing certain video games. For example, a user who tends to play video games for one or two play sessions and then move on to another video game may have a high churn rate. By identifying such users, it may be possible to modify the difficulty settings of the video game to reduce the rate of churn. For example, if it is determined that simple games, or games that do not provide an adequate challenge for the user results in the user ceasing to play such games, the user may be associated with a cluster of users who prefer difficult video games. Further, as is described in more detail below, based on the identity of the cluster to associate with the user, the difficulty of the video game112may be refined in an effort to reduce the probability that the user ceases to play the video game112.

Example Difficulty Setting Process

FIG. 6presents a flowchart of an embodiment of a difficulty setting process600for an application, such as the video game112. The process600can be implemented by any system that can dynamically set or adjust the difficulty of a video game based at least in part on monitor activity of the user. For example, the process600, in whole or in part, can be implemented by an interactive computing system130, a difficulty configuration system132, a user clustering system134, or a user computing system110, among others. Although any number of systems, in whole or in part, can implement the process600, to simplify discussion, the process600will be described with respect to particular systems.

The process600may be performed when the video game112is executed or when a play session of the video game112is started. In some cases, the process600may be performed when a user begins a new game or starts a new account with respect to video game112. Further, the process600may be performed each time the user loads a previously saved game with respect to the video game112. In some cases, the process600may be performed, or repeated, at a particular time or in response to a trigger event within the video game112. For example, the process600may occur each time the user fails to complete or succeeds in completing particular objectives within the video game112. In some cases, the process600may occur after a threshold number of objectives are completed or after a threshold number of attempts to complete an objective are unsuccessful.

The process600begins at block602where the difficulty configuration system132identifies a user of the video game112. In some cases, the block602is performed in response to the user accessing the video game112. However, as described above, one or more portions of the process600may be performed in response to other events, such as in game trigger events. Identifying the user of the video game112may be based on a user identifier or avatar. In some cases, the block602may include one or more of the embodiments previously described with respect to the block502.

At block604, the difficulty configuration system132determines a user cluster associated with the user. Determining a user cluster associated with the user may include accessing a user data repository138to identify one or more clusters associated with the user. In some cases, the user cluster may be associated with both the user and the video game112. In some such cases, the user may be associated with multiple user clusters. For example, the user may be associated with a different user cluster for each video game played by the user. In other cases, the user may be associated with a single user cluster for a set of video games. In some cases, the set of video games may be all video games played by the user or all video games published by an entity associated with the interactive computing system130. In other cases, the set of video games may be video games of a specific genre or theme.

Based on the identified user cluster, the difficulty configuration system132determines configuration values for a set of one or more knobs associated with the video game112. These configuration values may be identified by the user cluster. Alternatively, the configuration values may be determined based on a difficulty level associated with the user cluster. In some cases, at least some of the configuration values are specific or fixed values associated with the user cluster or determined based on characteristics associated with the user cluster. Alternatively, or in addition, at least some of the configuration values may be selected using an algorithm or randomly from a particular pool or set of values associated with the user cluster.

At block608, the difficulty configuration system132accesses recent user interaction data for the user with respect to the video game112. This recent user interaction data may be provided to the difficulty configuration system132by the video game112. Alternatively, or in addition, the difficulty configuration system132may access the recent user interaction data from the user interaction history repository116. In yet other cases, the difficulty configuration system132may access the recent user interaction data from the user data repository138.

The recent user interaction data may include user interaction data collected during a current play session, within a threshold time period (for example, within the last week), or within a threshold number of play sessions (for example, the most recent three play sessions). Generally, although not necessarily, the time period or the number of play sessions from which the recent user interaction data is collected is less and/or more recent than the historical data used with respect to the process500to determine a user cluster associated with the user.

Advantageously, in certain embodiments, by accessing the user cluster associated with the user and by accessing the recent user interaction data, the difficulty configuration system132can make a prediction as to the user's success at playing the video game112with a particular level of difficulty. Further, the difficulty configuration system132can make a prediction regarding the user's behavior with respect to the video game112. Based at least in part on these predictions, the difficulty configuration system132can select configuration values that can be sent to the video game112to adjust the difficulty of the video game112accordingly. In some embodiments, the block606or the block608may be optional or omitted. For example, in cases where the user has not previously played the video game112, the block608may be omitted. As another example, in cases where the user has not accumulated enough historical user interaction data to associate the user with a user cluster, the block606may be omitted. Alternatively, the user may be associated with a default user cluster so as to determine an initial set of configuration values for the set of knobs associated with the video game112. As another alternative, the user may be associated with a user cluster based at least in part on information provided by the user.

At block610, the difficulty configuration system132may adjust the configuration values for the set of knobs based at least in part on the recent user interaction data for the user. Adjusting the configuration values may make the video game112easier or more challenging for the user. Further, adjusting the configuration values may modify the state of the video game112and/or number of features associated with the video game112. For example, adjusting the configuration values may modify the layout of a level within the video game112. As another example, adjusting the configuration values may modify the timing of item drops within the video game112and/or the type of item drops within the video game112. In some cases, the configuration values may be seed values that are used by the video game112in generating one or more aspects of the video game112, such as a level layout or a start position of a user controlled character within the video game112. It should be understood that the present disclosure does not limit the type of modifications that may be made to the video game112. However, typically, the modifications made to the video game112will result in an adjustment of the level of difficulty of the video game112. Advantageously, in certain embodiments, by adjusting particular knobs within the video game112based on historical and/or recent user interaction data, the difficulty of the video game112may be adjusted in a more granular basis compared to static difficulty level settings associated with some video games.

In some cases, adjusting the difficulty settings for the video game may include changing configuration values that adjust a defined set of features in the video game112. Alternatively, the difficulty configuration system132may determine configuration settings based at least in part on an evaluation of specific skills of the user as determined from the recent user interaction data. For example, it may be determined that the user is failing a number of objectives in the video game112because the user has trouble timing jumps. If the user is a type of user who will stop playing a video game112when the user has trouble completing objectives, as determined by the user cluster associated with the user, the difficult configuration system132may generate configuration values for the capabilities of the playable character in the video game112to make it easier to time the jumps (for example, by allowing the character to jump farther). In contrast, if another user does not have a problem timing jumps, but has trouble aiming a weapon, the difficulty configuration system132may generate configuration values that make it easier to shoot objectives by making the hit area of a weapon larger. In some cases, both of these changes may be made without the users realizing that the game has been modified.

In some cases, the block610may include the difficulty configuration system132providing the video game112and/or the user computer system110the adjusted configuration values for the set of knobs. Providing the adjusted configuration values to the user computing system110may include providing the adjusted configuration values over a communication channel established between the interactive computing system130and the user computing system110. In some cases, communication channel may be between the video game112and the difficulty configuration system132. In some cases, the adjusted configuration values may be provided as a configuration file (for example, a markup language file, such as an XML file) that may be accessed by the video game112and/or that may override a configuration file of the video game112. In some cases, the adjuster configuration values may be provided in a shared library file such as a dynamic link library (DLL) file.

Although the process600may be used to provide configuration values that adjust the difficulty setting of the video game112, in some cases, the adjustment to the difficulty level of the video game112is bounded. For example, in some cases, the developer of the video game112may desire for each odd level to be easier than each even level within the video game112. Thus, the player will alternate between easier and harder levels while playing the video game112. In such a case, the adjustment to the configuration values at the block610may be restricted such that the odd levels remain easier than the even levels within the video game112. As another example, adjustments to the difficulty level of different portions of the video game112may be restricted to particular range. Thus, in some cases, although later levels in the video game112may be made more or less difficult compared to earlier levels, the later levels may still remain more difficult than the earlier levels.

In some embodiments, the process600may be performed at least in part by the user computing system110. For example, in cases where the user computing system110does not have a network connection to the interactive computing system130, the video game112may be configured to dynamically adjust its difficulty based at least in part on user interaction data stored in the user interaction history repository116. Further, in some cases, the interactive computing system130may provide user cluster information to the user computing system110for storage enabling the user computing system110to perform the process600at times when the user computing system110does not have a network connection to the interactive computing system130.

Example Seed Evaluation Process

FIG. 7presents a flowchart of an embodiment of a seed evaluation process700. The process700can be implemented by any system that can evaluate the difficulty of a portion of the video game112based at least in part on the use of a particular seed value with respect to the portion of the video game112. For example, the process700, in whole or in part, can be implemented by an interactive computing system130, a seed evaluation system136, a difficulty configuration system132, or a user computing system110, among others. Although any number of systems, in whole or in part, can implement the process700, to simplify discussion, the process700will be described with respect to particular systems.

The seed evaluated using the process700can relate to the generation of a number of aspects or portions of the video game112. For example, the seed may be used to determine an initial configuration of an in-game world or level. As another example, the seed may be used to determine the abilities of one or more playable or non-playable characters within the video game112.

The process700begins at block702where the seed evaluation system136identifies a number of seeds for configuring a video game112or a portion of the video game112. In some cases, each of the number of seeds may be associated with configuring the same portion of the video game112.

For each seed identified at the block702, the seed evaluation system136monitors the progress of the set of users playing or accessing the video game112over time to obtain progress data associated with the seed at block704. In some cases, each seed will be evaluated with a different set of users because, for example, only one seed may be used during a particular play session with a particular account of a particular user for some types of video games. For example, if the seed value is associated with a layout of a particular level, it is possible that the layout of the particular level will not change once set for a particular user. However, in some other cases, each seed may be evaluated with the same set of users because, for example, the seed values are used repeatedly during play sessions of the video game. For example, if the seed value is associated with a starting set of cards in a card battle game, a starting set of cards may be generated each time the user plays a round providing an opportunity for a number of different seed values to be evaluated with a particular user.

In some implementations, the set of users monitored at the block704are selected based on the skill level associated with the set of users or a user cluster associated with a set of users. For example, each of the seed values may be evaluated by users who are determined to have roughly the same skill level. Alternatively, each of the seed values may be evaluated by a number of users with varying degrees of skill.

The progress data obtained at the block704may reflect the amount of progress that the monitored users have made playing the video game112with a particular seed value. For example, the progress data may indicate whether a particular user or set of users completed an objective with a particular seed value or failed to complete an objective with the particular seed value.

At block706, for each seed evaluated, the seed evaluation system136normalizes the progress data based on skill data associated with each user from each of the sets of users. In other words, in some cases, the progress data obtained for each of the seed values may be normalized so that the different skill levels of different users do not impact the evaluation of the seeds. Alternatively, or in addition, progress data obtained with respect to a particular user for a seed may be weighted based on the user's skill level as determined by, for example, the user clustering system134. In some cases, the user's skill level may be determined based on points earned playing the video game112or some other metric for monitoring the skills of a particular user.

In some embodiments, the block706may be optional or omitted. For example, in cases where the set of users selected to evaluate each seed are associated with a particular user cluster or are each determined to have less than a threshold difference of skill level, it may be unnecessary to normalize the progress data and the block706may be omitted.

At block708, the seed evaluation system136determines a difficulty associated with each seed based on the associated normalize progress data. The difficulty may be determined based on the number or percentage of users who are successful or unsuccessful in completing an objective or a portion of the video game112when a particular seed value associated with the objective or the portion of the video game112is used. Advantageously, in certain embodiments, by evaluating the difficulty of seed values, it is possible to group or cluster sets of seed values for a particular difficulty levels. Thus, in some cases, if it is determined that a particular user prefers to play easier video games, seed values that are associated with a lower difficulty level may be used to generate portions of the video game112. Further, the process700enables developers to confirm the level of difficulty associated with a particular seed value and to make adjustments to the classification of a particular seed value. For example, a particular seed value may be used with players who select an easy difficulty level. However, after evaluating a number of play sessions for a number of users with the particular seed value using the process700, it may be determined that the seed value causes a portion of the video game112to be significantly more challenging than when the portion of the video game112is associated with other seed values. In such a case, the particular seed value may be reclassified for use with players who prefer or select a harder difficulty level and may be removed from availability for use with players who prefer or select an easy difficulty level.

Overview of Computing System

FIG. 8illustrates an embodiment of a user computing system110, which may also be referred to as a gaming system. As illustrated, the user computing system110may be a single computing device that can include a number of elements. However, in some cases, the user computing system110may include multiple devices. For example, the user computing system110may include one device that includes that includes a central processing unit and a graphics processing unit, another device that includes a display, and another device that includes an input mechanism, such as a keyboard or mouse.

The user computing system110can be an embodiment of a computing system that can execute a game system. In the non-limiting example ofFIG. 8, the user computing system110is a touch-capable computing device capable of receiving input from a user via a touchscreen display802. However, the user computing system110is not limited as such and may include non-touch capable embodiments, which do not include a touchscreen display802.

The user computing system110includes a touchscreen display802and a touchscreen interface804, and is configured to execute a game application810. This game application may be the video game112or an application that executes in conjunction with or in support of the video game112, such as a video game execution environment. Although described as a game application810, in some embodiments the application810may be another type of application that may have a variable execution state based at least in part on the preferences or capabilities of a user, such as educational software. While user computing system110includes the touchscreen display802, it is recognized that a variety of input devices may be used in addition to or in place of the touchscreen display802.

The user computing system110can include one or more processors, such as central processing units (CPUs), graphics processing units (GPUs), and accelerated processing units (APUs). Further, the user computing system110may include one or more data storage elements. In some embodiments, the user computing system110can be a specialized computing device created for the purpose of executing game applications810. For example, the user computing system110may be a video game console. The game applications810executed by the user computing system110may be created using a particular application programming interface (API) or compiled into a particular instruction set that may be specific to the user computing system110. In some embodiments, the user computing system110may be a general purpose computing device capable of executing game applications810and non-game applications. For example, the user computing system110may be a laptop with an integrated touchscreen display or desktop computer with an external touchscreen display. Components of an example embodiment of a user computing system110are described in more detail with respect toFIG. 9.

The touchscreen display802can be a capacitive touchscreen, a resistive touchscreen, a surface acoustic wave touchscreen, or other type of touchscreen technology that is configured to receive tactile inputs, also referred to as touch inputs, from a user. For example, the touch inputs can be received via a finger touching the screen, multiple fingers touching the screen, a stylus, or other stimuli that can be used to register a touch input on the touchscreen display802. The touchscreen interface804can be configured to translate the touch input into data and output the data such that it can be interpreted by components of the user computing system110, such as an operating system and the game application810. The touchscreen interface804can translate characteristics of the tactile touch input touch into touch input data. Some example characteristics of a touch input can include, shape, size, pressure, location, direction, momentum, duration, and/or other characteristics. The touchscreen interface804can be configured to determine the type of touch input, such as, for example a tap (for example, touch and release at a single location) or a swipe (for example, movement through a plurality of locations on touchscreen in a single touch input). The touchscreen interface804can be configured to detect and output touch input data associated with multiple touch inputs occurring simultaneously or substantially in parallel. In some cases, the simultaneous touch inputs may include instances where a user maintains a first touch on the touchscreen display802while subsequently performing a second touch on the touchscreen display802. The touchscreen interface804can be configured to detect movement of the touch inputs. The touch input data can be transmitted to components of the user computing system110for processing. For example, the touch input data can be transmitted directly to the game application810for processing.

In some embodiments, the touch input data can undergo processing and/or filtering by the touchscreen interface804, an operating system, or other components prior to being output to the game application810. As one example, raw touch input data can be captured from a touch input. The raw data can be filtered to remove background noise, pressure values associated with the input can be measured, and location coordinates associated with the touch input can be calculated. The type of touch input data provided to the game application810can be dependent upon the specific implementation of the touchscreen interface804and the particular API associated with the touchscreen interface804. In some embodiments, the touch input data can include location coordinates of the touch input. The touch signal data can be output at a defined frequency. Processing the touch inputs can be computed many times per second and the touch input data can be output to the game application for further processing.

A game application810can be configured to be executed on the user computing system110. The game application810may also be referred to as a video game, a game, game code and/or a game program. A game application should be understood to include software code that a user computing system110can use to provide a game for a user to play. A game application810might comprise software code that informs a user computing system110of processor instructions to execute, but might also include data used in the playing of the game, such as data relating to constants, images and other data structures. For example, in the illustrated embodiment, the game application includes a game engine812, game data814, and game state information816.

The touchscreen interface804or another component of the user computing system110, such as the operating system, can provide user input, such as touch inputs, to the game application810. In some embodiments, the user computing system110may include alternative or additional user input devices, such as a mouse, a keyboard, a camera, a game controller, and the like. A user can interact with the game application810via the touchscreen interface804and/or one or more of the alternative or additional user input devices. The game engine812can be configured to execute aspects of the operation of the game application810within the user computing system110. Execution of aspects of gameplay within a game application can be based, at least in part, on the user input received, the game data814, and game state information816. The game data814can include game rules, prerecorded motion capture poses/paths, environmental settings, constraints, animation reference curves, skeleton models, and/or other game application information. Further, the game data814may include information that is used to set or adjust the difficulty of the game application810.

The game engine812can execute gameplay within the game according to the game rules. Some examples of game rules can include rules for scoring, possible inputs, actions/events, movement in response to inputs, and the like. Other components can control what inputs are accepted and how the game progresses, and other aspects of gameplay. During execution of the game application810, the game application810can store game state information816, which can include character states, environment states, scene object storage, and/or other information associated with a state of execution of the game application810. For example, the game state information816can identify the state of the game application at a specific point in time, such as a character position, character action, game level attributes, and other information contributing to a state of the game application.

The game engine812can receive the user inputs and determine in-game events, such as actions, collisions, runs, throws, attacks and other events appropriate for the game application810. During operation, the game engine812can read in game data814and game state information816in order to determine the appropriate in-game events. In one example, after the game engine812determines the character events, the character events can be conveyed to a movement engine that can determine the appropriate motions the characters should make in response to the events and passes those motions on to an animation engine. The animation engine can determine new poses for the characters and provide the new poses to a skinning and rendering engine. The skinning and rendering engine, in turn, can provide character images to an object combiner in order to combine animate, inanimate, and background objects into a full scene. The full scene can conveyed to a renderer, which can generate a new frame for display to the user. The process can be repeated for rendering each frame during execution of the game application. Though the process has been described in the context of a character, the process can be applied to any process for processing events and rendering the output for display to a user.

Example Hardware Configuration of Computing System

FIG. 9illustrates an embodiment of a hardware configuration for the user computing system110ofFIG. 8. Other variations of the user computing system110may be substituted for the examples explicitly presented herein, such as removing or adding components to the user computing system110. The user computing system110may include a dedicated game device, a smart phone, a tablet, a personal computer, a desktop, a laptop, a smart television, a car console display, and the like. Further, (although not explicitly illustrated inFIG. 9) as described with respect toFIG. 8, the user computing system110may optionally include a touchscreen display802and a touchscreen interface804.

As shown, the user computing system110includes a processing unit20that interacts with other components of the user computing system110and also components external to the user computing system110. A game media reader22may be included that can communicate with game media12. Game media reader22may be an optical disc reader capable of reading optical discs, such as CD-ROM or DVDs, or any other type of reader that can receive and read data from game media12. In some embodiments, the game media reader22may be optional or omitted. For example, game content or applications may be accessed over a network via the network I/O38rendering the game media reader22and/or the game media12optional.

The user computing system110may include a separate graphics processor24. In some cases, the graphics processor24may be built into the processing unit20, such as with an APU. In some such cases, the graphics processor24may share Random Access Memory (RAM) with the processing unit20. Alternatively, or in addition, the user computing system110may include a discrete graphics processor24that is separate from the processing unit20. In some such cases, the graphics processor24may have separate RAM from the processing unit20. Further, in some cases, the graphics processor24may work in conjunction with one or more additional graphics processors and/or with an embedded or non-discrete graphics processing unit, which may be embedded into a motherboard and which is sometimes referred to as an on-board graphics chip or device.

The user computing system110also includes various components for enabling input/output, such as an I/O32, a user I/O34, a display I/O36, and a network I/O38. As previously described, the input/output components may, in some cases, including touch-enabled devices. The I/O32interacts with storage element40and, through a device42, removable storage media44in order to provide storage for computing device800. Processing unit20can communicate through I/O32to store data, such as game state data and any shared data files. In addition to storage40and removable storage media44, computing device800is also shown including ROM (Read-Only Memory)46and RAM48. RAM48may be used for data that is accessed frequently, such as when a game is being played.

User I/O34is used to send and receive commands between processing unit20and user devices, such as game controllers. In some embodiments, the user I/O34can include touchscreen inputs. As previously described, the touchscreen can be a capacitive touchscreen, a resistive touchscreen, or other type of touchscreen technology that is configured to receive user input through tactile inputs from the user. Display I/O36provides input/output functions that are used to display images from the game being played. Network I/O38is used for input/output functions for a network. Network I/O38may be used during execution of a game, such as when a game is being played online or being accessed online.

Display output signals may be produced by the display I/O36and can include signals for displaying visual content produced by the computing device800on a display device, such as graphics, user interfaces, video, and/or other visual content. The user computing system110may comprise one or more integrated displays configured to receive display output signals produced by the display I/O36, which may be output for display to a user. According to some embodiments, display output signals produced by the display I/O36may also be output to one or more display devices external to the computing device800.

The user computing system110can also include other features that may be used with a game, such as a clock50, flash memory52, and other components. An audio/video player56might also be used to play a video sequence, such as a movie. It should be understood that other components may be provided in the user computing system110and that a person skilled in the art will appreciate other variations of the user computing system110.

Program code can be stored in ROM46, RAM48, or storage40(which might comprise hard disk, other magnetic storage, optical storage, solid state drives, and/or other non-volatile storage, or a combination or variation of these). At least part of the program code can be stored in ROM that is programmable (ROM, PROM, EPROM, EEPROM, and so forth), in storage40, and/or on removable media such as game media12(which can be a CD-ROM, cartridge, memory chip or the like, or obtained over a network or other electronic channel as needed). In general, program code can be found embodied in a tangible non-transitory signal-bearing medium.

Random access memory (RAM)48(and possibly other storage) is usable to store variables and other game and processor data as needed. RAM is used and holds data that is generated during the play of the game and portions thereof might also be reserved for frame buffers, game state and/or other data needed or usable for interpreting user input and generating game displays. Generally, RAM48is volatile storage and data stored within RAM48may be lost when the user computing system110is turned off or loses power.

As user computing system110reads game media12and provides a game, information may be read from game media12and stored in a memory device, such as RAM48. Additionally, data from storage40, ROM46, servers accessed via a network (not shown), or removable storage media46may be read and loaded into RAM48. Although data is described as being found in RAM48, it will be understood that data does not have to be stored in RAM48and may be stored in other memory accessible to processing unit20or distributed among several media, such as game media12and storage40.

Additional Embodiments

In certain embodiments, a computer-implemented method is disclosed that may be implemented by an interactive computing system configured with specific computer-executable instructions to at least determine a user identifier of a first user who is playing a video game on a user computing device. The method may further include identifying a user cluster associated with the first user from a plurality of user clusters based at least in part on the user identifier of the first user. Each user cluster from the plurality of user clusters may correspond to different state preferences for the video game. Moreover, based at least in part on state preferences associated with the identified user cluster, the method may include determining a configuration value for a knob associated with the video game. The knob may include a variable that when adjusted causes a modification to a state of the video game. In addition, the method may include accessing recent user interaction data associated with the first user playing the video game over a first time period and determining an adjustment to the configuration value based at least in part on the recent user interaction data to obtain a modified configuration value. Further, the method may include generating a configuration data package that comprises the modified configuration value for transmission to the user computing device for modifying execution of the video game by adjusting the knob based at least in part on the modified configuration value.

In some implementations, the modification to the state of the video game adjusts a difficulty of the video game. Further, the first time period may be less than a first threshold time period. In addition, the method may include generating the plurality of user clusters by at least identifying a plurality of users who play a second video game and obtaining user interaction data for each user of the plurality of users by at least monitoring the user's interaction with the second video game over a second time period. In addition, the method may include determining a level of engagement for each user of the plurality of users based at least in part on the user interaction data and determining the plurality of user clusters based at least in part on the user interaction data for each user of the plurality of users and the level of engagement for each user of the plurality of users. In some cases, the second video game and the video game are the same. Further, the second time period may be greater than a second threshold time period.

In some embodiments, the method includes associating the first user with the user cluster by at least obtaining historical user interaction data for the first user by at least monitoring the first user's interaction with a third video game over a third time period, accessing cluster definitions for the plurality of user clusters, and selecting the user cluster from the plurality of user clusters based at least in part on the historical user interaction data and the cluster definitions for the plurality of user clusters. In some cases, the third video game and the video game are the same. Further, the third time period may be longer than the first time period.

In certain embodiments of the present disclosure, a system comprising an electronic data store configured to store user interaction data with respect to a video game and a hardware processor in communication with the electronic data store is disclosed. The hardware processor may be configured to execute specific computer-executable instructions to at least identify a user cluster associated with a first user from a plurality of user clusters. At least some of the user clusters from the plurality of user clusters may correspond to different state preferences for video games played by users associated with the user cluster. The system, based at least in part on state preferences associated with the identified user cluster, may further determine a configuration value for a state variable associated with a video game played by the first user. In addition, the system may access, from the electronic data store, recent user interaction data associated with the first user accessing the video game over a first time period and determine an adjustment to the configuration value based at least in part on the recent user interaction data to obtain a modified configuration value. Moreover, the system may generate a configuration data package that comprises the modified configuration value for transmission to a user computing device for modifying execution of the video game by adjusting the state variable based at least in part on the modified configuration value.

In certain implementations, the modified execution of the video game is undetectable by the first user. Further, the first time period may include a dynamic time window that adjusts with the passage of time. In some embodiments, at least the accessing recent user interaction data, the determining an adjustment to the configuration value, and the modifying the execution of the video game is repeated in response to a trigger event occurring during a play session of the video game by the first user.

In some implementations, the hardware processor is further configured to generate the plurality of user clusters by executing specific computer-executable instructions to at least identify a plurality of users who play the video game and access user interaction data associated with accessing the video game from the electronic data store for each user of the plurality of users. The user interaction data may be recorded over a second time period. Further, the system may determine a level of engagement for each user of the plurality of users based at least in part on the user interaction data and determine the plurality of user clusters based at least in part on the user interaction data for each user of the plurality of users and the level of engagement for each user of the plurality of users.

Further, the system may associate the first user with the user cluster by executing specific computer-executable instructions to at least obtain historical user interaction data for the first user by at least monitoring the first user's access of the video game over a third time period. The system may further access cluster definitions for the plurality of user clusters and select the user cluster from the plurality of user clusters based at least in part on the historical user interaction data and the cluster definitions for the plurality of clusters. In some cases, the third time period precedes the first time period and the third time period is longer than the first time period.

In certain embodiments of the present disclosure, a non-transitory computer-readable storage medium storing computer executable instructions that, when executed by one or more computing devices, configure the one or more computing devices to perform operations comprising identifying a user cluster associated with a first user from a plurality of user clusters. At least some of the user clusters from the plurality of user clusters may correspond to different state settings for applications accessed by users associated with the user cluster. Moreover, based at least in part on state settings associated with the identified user cluster, the operations may include determining a configuration value for a state variable associated with an application accessed by the first user and accessing recent user interaction data associated with the first user accessing the application over a first time period. In addition, the operations may include determining an adjustment to the configuration value based at least in part on the recent user interaction data to obtain a modified configuration value and generating a configuration data package that comprises the modified configuration value for transmission to a user computing device for modifying execution of the application by adjusting the state variable based at least in part on the modified configuration value.

In some implementations, the operations further comprise generating the plurality of user clusters by at least identifying a plurality of users who access the application and obtaining user interaction data for each user of the plurality of users by at least monitoring the user's interaction with the application over a second time period. In addition, the operations may include determining a level of engagement for each user of the plurality of users based at least in part on the user interaction data and determining the plurality of user clusters based at least in part on the user interaction data for each user of the plurality of users and the level of engagement for each user of the plurality of users.

In some cases, the operations may further include associating the first user with the user cluster by at least obtaining historical user interaction data for the first user by at least monitoring the first user's access of the application over a third time period and determining difficulty preferences for the first user based at least in part on the historical user interaction data. In addition, the operations may include accessing cluster definitions for the plurality of user clusters. Each cluster definition may identify difficulty preferences for users associated with the corresponding user cluster. Further, the operations may include selecting the user cluster from the plurality of user clusters by matching the difficulty preference for the first user with the difficulty preferences of the cluster definition corresponding to the user cluster. In addition, the third time period may begin at a point in time that precedes the start of the first time period and the third time period may be longer than the first time period.

It is to be understood that not necessarily all objects or advantages may be achieved in accordance with any particular embodiment described herein. Thus, for example, those skilled in the art will recognize that certain embodiments may be configured to operate in a manner that achieves or optimizes one advantage or group of advantages as taught herein without necessarily achieving other objects or advantages as may be taught or suggested herein.

All of the processes described herein may be embodied in, and fully automated via, software code modules executed by a computing system that includes one or more computers or processors. The code modules may be stored in any type of non-transitory computer-readable medium or other computer storage device. Some or all the methods may be embodied in specialized computer hardware.

Many other variations than those described herein will be apparent from this disclosure. For example, depending on the embodiment, certain acts, events, or functions of any of the algorithms described herein can be performed in a different sequence, can be added, merged, or left out altogether (for example, not all described acts or events are necessary for the practice of the algorithms). Moreover, in certain embodiments, acts or events can be performed concurrently, for example, through multi-threaded processing, interrupt processing, or multiple processors or processor cores or on other parallel architectures, rather than sequentially. In addition, different tasks or processes can be performed by different machines and/or computing systems that can function together.

The various illustrative logical blocks and modules described in connection with the embodiments disclosed herein can be implemented or performed by a machine, such as a processing unit or processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A processor can be a microprocessor, but in the alternative, the processor can be a controller, microcontroller, or state machine, combinations of the same, or the like. A processor can include electrical circuitry configured to process computer-executable instructions. In another embodiment, a processor includes an FPGA or other programmable device that performs logic operations without processing computer-executable instructions. A processor can also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration. Although described herein primarily with respect to digital technology, a processor may also include primarily analog components. A computing environment can include any type of computer system, including, but not limited to, a computer system based on a microprocessor, a mainframe computer, a digital signal processor, a portable computing device, a device controller, or a computational engine within an appliance, to name a few.

Conditional language such as, among others, “can,” “could,” “might” or “may,” unless specifically stated otherwise, are otherwise understood within the context as used in general to convey that certain embodiments include, while other embodiments do not include, certain features, elements and/or steps. Thus, such conditional language is not generally intended to imply that features, elements and/or steps are in any way required for one or more embodiments or that one or more embodiments necessarily include logic for deciding, with or without user input or prompting, whether these features, elements and/or steps are included or are to be performed in any particular embodiment.

Disjunctive language such as the phrase “at least one of X, Y, or Z,” unless specifically stated otherwise, is otherwise understood with the context as used in general to present that an item, term, etc., may be either X, Y, or Z, or any combination thereof (for example, X, Y, and/or Z). Thus, such disjunctive language is not generally intended to, and should not, imply that certain embodiments require at least one of X, at least one of Y, or at least one of Z to each be present.

Any process descriptions, elements or blocks in the flow diagrams described herein and/or depicted in the attached figures should be understood as potentially representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or elements in the process. Alternate implementations are included within the scope of the embodiments described herein in which elements or functions may be deleted, executed out of order from that shown, or discussed, including substantially concurrently or in reverse order, depending on the functionality involved as would be understood by those skilled in the art.

Unless otherwise explicitly stated, articles such as “a” or “an” should generally be interpreted to include one or more described items. Accordingly, phrases such as “a device configured to” are intended to include one or more recited devices. Such one or more recited devices can also be collectively configured to carry out the stated recitations. For example, “a processor configured to carry out recitations A, B and C” can include a first processor configured to carry out recitation A working in conjunction with a second processor configured to carry out recitations B and C.

It should be emphasized that many variations and modifications may be made to the above-described embodiments, the elements of which are to be understood as being among other acceptable examples. All such modifications and variations are intended to be included herein within the scope of this disclosure.