A game behavior analysis system based on deep learning

Abstract

The present invention relates to the field of behavioral analysis technology, specifically a deep learning-based gaming behavior analysis system, the system comprising a gaming activity analysis module, a gaming social behavior analysis module, a user preference identification module, and an update impact analysis module. The present invention utilizes user gaming behavior data to comprehensively assess users' daily activity patterns and long-term engagement trends, providing a deeper and more accurate understanding of user behavior. By integrating user behavioral data and social behavior results, the system is able to assess the importance of each user, identify important gaming users, and identify key gaming content based on the user's importance and gaming behavior data. Subsequently, in subsequent updates, updates can be made targeting key user-preferred content, improving user satisfaction and loyalty. Furthermore, after the game is updated, the system can determine the actual effects of the update, providing a scientific basis for future game design and optimization.

Description

Technical Field

[0001] The present invention relates to the field of behavior analysis technology, and in particular to a game behavior analysis system based on deep learning. Background Art

[0002] The field of behavioral analytics encompasses a broad range of methods and techniques used to study, evaluate, and predict the behavior of individuals or groups. This field combines theories and techniques from psychology, sociology, human-computer interaction, and computer science. By employing machine learning, statistical analysis, pattern recognition, and algorithm optimization, behavioral analytics techniques can extract meaningful behavioral patterns and trends from complex data. This technology is widely used in areas such as market research, user experience design, security monitoring, and health management, helping organizations and businesses better understand customer or user behavior and thus optimize product design and service delivery.

[0003] The Game Behavior Analysis System monitors and analyzes player behavior patterns within video games. The system's primary purpose is to capture and analyze player interaction data to understand player behavior, preferences, and evolving trends. This information can be used to iteratively optimize game design, enhancing player experience and satisfaction. Furthermore, through a deeper understanding of player behavior, developers can effectively adjust game mechanics to increase player engagement and loyalty.

[0004] Traditional systems focus on collecting and simply analyzing static data, failing to deeply explore the behavioral patterns and user needs underlying this data. This prevents the analysis results from being directly applied to actual improvements in game design, limiting the responsiveness and innovation of game development. Traditional systems overlook subtle shifts in user behavior, such as changes in the frequency and content of in-game conversations. Lacking analytical tools sensitive to dynamic changes, these systems hinder developers from quickly identifying and responding to changes in user behavior, leading to user churn and lost market opportunities. Traditional systems also lack analysis of user behavior before and after game updates, a particularly significant issue in the gaming industry, which requires rapid feedback and adjustments. This leads to delayed feedback on player behavior, impacting the overall player experience and satisfaction. Summary of the Invention

[0005] The purpose of this invention is to solve the shortcomings of the existing technology and propose a game behavior analysis system based on deep learning.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a game behavior analysis system based on deep learning, the system comprising:

[0007] The game activity analysis module counts users' daily play time based on their gaming behavior data, analyzes their average weekly online times, evaluates their activity, analyzes their activity data, evaluates their stability, and obtains user activity analysis results.

[0008] The game social behavior analysis module uses a deep learning model to analyze user conversation behavior data in the game, identify user conversation types, and evaluate the impact of user communication behavior in the game. Combined with the number of friends a user has in the game, it evaluates the user's social influence in the game and obtains the user's social behavior analysis results.

[0009] The user preference identification module evaluates the importance level of each user based on the user's activity, stability, and social influence based on the user's activity and social behavior analysis results, extracts the user's behavior data, evaluates the overall user's preference for differentiated game content, identifies the user's preferred game content, and obtains a user preference content identification result;

[0010] The update impact analysis module extracts the changes in user activity and stability before and after the game update based on the user preference content identification results, combines the changes in the frequency and time of users playing differentiated game content, evaluates the effect of the game update, and obtains the game update impact analysis results.

[0011] The present invention is improved in that the step of evaluating the user's activity level is as follows:

[0012] Based on the user's gaming behavior data, the user's playing time is counted, and the user's average weekly online times are analyzed to obtain activity-related data;

[0013] Based on the activity-related data, the formula:

[0014] ;

[0015] Calculate the user's activity index;

[0016] in, The user's daily play time, The total number of times the user has been online in the past week. and is the weight coefficient, is the user's activity index;

[0017] Based on the activity index of the user, the activity of the user is evaluated according to the size of the activity index to obtain user activity information.

[0018] The present invention is improved in that the steps of obtaining the user activity analysis results are as follows:

[0019] Based on the user activity information, extracting user activity data within a period of time according to a preset time interval to obtain stability correlation information;

[0020] Based on the stability association information, the formula:

[0021] ;

[0022] Calculate the user's stability index;

[0023] in, is the user's stability index, is the total number of days in the statistical period, For the User activity index for the day, is the average activity index;

[0024] Based on the user's stability index, the user's stability is evaluated according to the size of the stability index, and the user's activity index is integrated to obtain a user activity analysis result.

[0025] The present invention is improved in that the step of evaluating the impact of the user's communication behavior in the game is:

[0026] Based on the user's in-game conversation behavior data, a deep learning model is used to extract keywords from the user's conversation information, analyze the user's in-game conversation information, and identify the user's conversation type, including normal game communication, normal life conversation, and abnormal conversation, to obtain the user conversation type recognition result;

[0027] Based on the user conversation type identification results, the frequency of each user conversation type is counted to obtain communication impact correlation data;

[0028] Based on the communication impact correlation data, the formula:

[0029] ;

[0030] Calculate the user's communication impact index and obtain the user's communication impact assessment result;

[0031] in, The user's normal game communication frequency, The frequency of normal life conversations for users, is the user's abnormal conversation frequency, 、 and is the impact factor, The communication impact index of the user.

[0032] The present invention is improved in that the steps of obtaining the user social behavior analysis results are as follows:

[0033] Based on the user communication impact evaluation results, extract the number of game friends of the user and calculate the average number of friends of all users to obtain social influence correlation data;

[0034] Based on the social influence correlation data, the formula:

[0035] ;

[0036] Calculate the user's social influence index and obtain the user's social behavior analysis results;

[0037] in, is the average number of friends of game users, is the user's communication impact index, is the number of friends of the user, is the average activity of the user’s friends, The social influence index of the user.

[0038] The present invention is improved in that the step of evaluating the importance level of each user is as follows:

[0039] Based on the user activity analysis results and the user social behavior analysis results, extract the user's activity index, stability index and social influence index, and extract the preset minimum importance level to obtain importance-related data;

[0040] Based on the importance associated data, the formula:

[0041] ;

[0042] Calculate the user's importance level and obtain the user's importance evaluation result;

[0043] in, is the preset minimum importance level, is the user's activity index, is the user's stability index, is the user's social influence index, 、 and is the weight coefficient, The importance level of the user.

[0044] The present invention is improved in that the steps of obtaining the user preference content identification result are:

[0045] Based on the user importance evaluation results, by counting the number of users with differentiated importance levels, analyzing the proportion of users with differentiated importance levels, and extracting behavioral data of users with differentiated importance levels, including the frequency and time of playing each type of game content, to obtain game preference association data;

[0046] Based on the game preference association data, the formula:

[0047] ;

[0048] Calculate the user's preference index for game content;

[0049] in, For the The proportion of users with different importance levels, For the The number of times users of different importance levels participate in game content, For the The total playing time of users with different levels of importance on game content, For the The value of the importance level, is the total number of importance levels, The user's preference index for game content;

[0050] Based on the user's preference index for game content, the game content is sorted according to the size of the preference index, the game content preferred by the user is identified, and a user preference content identification result is obtained.

[0051] The present invention is improved in that the steps for obtaining the game update impact analysis results are as follows:

[0052] Based on the user preference content identification results, implement game updates according to the user's preference content, and extract user activity index and stability index data before and after the update, and extract changes in user participation in the target game content, including changes in play frequency and time, to obtain update effect correlation data;

[0053] Based on the update effect associated data, by the formula:

[0054] ;

[0055] Calculate the game update effect score;

[0056] in, Rate the game update performance, is the change in activity index, is the change in stability index, The change in frequency of game participation for target update content, The amount of game time change for the target update content, 、 and is the weight coefficient, is a symbolic function that returns The symbol, when When is a positive number, For the right, when When it is a negative number, is negative;

[0057] Based on the game update effect score, the game update effect is evaluated according to the positive or negative and size of the score, and it is determined whether the game update effect meets expectations, thereby obtaining a game update impact analysis result.

[0058] Compared with the prior art, the advantages and positive effects of the present invention are:

[0059] In the present invention, by utilizing user game behavior data, users' daily active patterns and long-term participation trends are comprehensively evaluated, and the understanding of user behavior is more in-depth and accurate. Through in-depth analysis of in-game conversation content, user communication patterns and their social influence are identified. By integrating user behavior data and social behavior results, the importance of each user can be evaluated, important game users can be identified, and key game content can be identified based on the user's importance and game behavior data. In subsequent updates, key content of user preferences can be updated to improve user satisfaction and loyalty. After the game is updated, a before-and-after comparative analysis of activity and stability, as well as an evaluation of changes in participation, can accurately judge the actual effect of the update, providing a scientific basis for future game design and optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a system flow chart of the present invention;

[0061] Figure 2 A flowchart for evaluating user activity in the present invention;

[0062] Figure 3 This is a flow chart of the present invention for obtaining user activity analysis results;

[0063] Figure 4 A flowchart for evaluating the impact of a user's communication behavior in a game according to the present invention;

[0064] Figure 5 This is a flow chart of the user social behavior analysis results of the present invention;

[0065] Figure 6 A flow chart for evaluating the importance level of each user for the present invention;

[0066] Figure 7 This is a flow chart of the present invention for obtaining the result of identifying the user's favorite content;

[0067] Figure 8 This is a flowchart of the present invention for obtaining game update impact analysis results. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0069] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description. They do not indicate or imply that the devices or elements referred to must have a specific direction, be constructed and operate in a specific direction, and therefore should not be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0070] See also Figure 1 The present invention provides a technical solution: a game behavior analysis system based on deep learning, the system comprising:

[0071] The game activity analysis module counts users' daily play time based on their gaming behavior data, analyzes their average weekly online times, evaluates their activity, analyzes their activity data, evaluates their stability, and obtains user activity analysis results.

[0072] The game social behavior analysis module uses a deep learning model to extract keywords from user conversation information based on in-game conversation data. It then analyzes the user's in-game conversation information and identifies the user's conversation types, including normal game communication, normal life conversations, and abnormal conversations. It then evaluates the impact of the user's in-game communication behavior based on the frequency of the user's in-game conversations and the number of friends the user has in the game to assess the user's social influence in the game, ultimately generating user social behavior analysis results.

[0073] The user preference identification module evaluates the importance level of each user based on the user's activity, stability, and social influence, based on the user's activity and social behavior analysis results. It also extracts user behavior data, including the frequency and time of playing differentiated game content. Combined with the proportion of users with differentiated importance levels, it evaluates the overall user preference for differentiated game content, identifies the user's preferred game content, and obtains the user preference content identification results;

[0074] The update impact analysis module is based on the results of user preference content identification, extracts changes in user activity and stability before and after the game update, and combines the changes in the frequency and time of users playing differentiated game content to evaluate the effect of the game update and obtain the game update impact analysis results.

[0075] The user activity analysis results include the average daily playing time, average weekly online times, activity score and stability score; the user social behavior analysis results include conversation type classification information, conversation frequency, social network density and social influence score; the user preference content identification results include game content preference, playing time information and frequency ranking information; the game update impact analysis results include activity change rate, stability change index, user retention rate and comparison of playing time before and after the update.

[0076] See also Figure 2 ,The steps to evaluate user activity are:

[0077] Based on the user's gaming behavior data, the user's playing time is counted, and the user's average weekly online times are analyzed to obtain activity-related data;

[0078] Based on the activity correlation data, the formula is:

[0079] ;

[0080] Calculate the user's activity index, is the user's activity index;

[0081] in, The user's daily play time, The total number of times the user has been online in the past week. and is the weight coefficient;

[0082] Based on the user's activity index, the user's activity is evaluated according to the size of the activity index to obtain user activity information.

[0083] formula:

[0084] ;

[0085] The meaning and acquisition method of the parameters:

[0086] : The user's daily play time, which is obtained by recording the total daily play time of the user.

[0087] : The total number of times a user logged in during the most recent week. This is calculated by recording a user's daily login times and adding them up. For example, if a user logged in 7, 9, 8, 10, 9, 11, and 12 times in a week, their total number of logins for the week would be 7+9+8+10+9+11+12=66.

[0088] and : Weight coefficient, used to adjust the contribution of daily play time and weekly online times in the activity index calculation. The selection of the weight coefficient relies on regression analysis of historical data, and the optimal value is usually determined by minimizing the prediction error.

[0089] Calculation example:

[0090] Assume a user's data: the user's daily playing time is 15 hours / day, the total number of online times in the past week is 66 times, and the weight coefficient determined by historical data analysis is and .

[0091] Activity Index The calculation formula is:

[0092] ;

[0093] Calculate the logarithmic portion of daily play time: ;

[0094] Calculate the square root of the number of times you go online in a week: ;

[0095] Substitute the above results into the formula: ;

[0096] The Activity Index indicates the user's activity level. A higher value indicates a more active user. This value is calculated by weighting the user's play time and login frequency, providing a more comprehensive reflection of the user's activity level.

[0097] See also Figure 3 , the steps to obtain the user activity analysis results are:

[0098] Based on user activity information, extract user activity data within a period of time according to a preset time interval to obtain stability correlation information;

[0099] Based on the stability correlation information, the formula:

[0100] ;

[0101] Calculate the user's stability index;

[0102] in, is the user's stability index, is the total number of days in the statistical period, For the User activity index for the day, is the average activity index;

[0103] Based on the user's stability index, the user's stability is evaluated according to the size of the stability index, and the user's activity index is integrated to obtain the user activity analysis results.

[0104] formula:

[0105] ;

[0106] Parameter details and how to obtain them:

[0107] It is the total number of days in the statistical period, such as the number of days in a month or six months, and is usually determined based on the available range of user activity data.

[0108] Is the user in The activity index of the day is calculated using the method in the previous step.

[0109] is the average activity index, expressed as , provides the average level of user activity and is used to measure the deviation of an individual from the average level.

[0110] Calculation example:

[0111] Set the activity index of a user within four days to be , , , .

[0112] Calculating average activity :

[0113] ;

[0114] Calculate the square of the difference between the daily activity index and the average, and then sum them:

[0115]

[0116]

[0117] ;

[0118] calculate :

[0119] ;

[0120] The calculated results indicate that the user's activity level has low volatility, indicating that the user's behavior is relatively stable over the observation period. A high stability index indicates that the user's daily activity pattern is relatively consistent and has low volatility.

[0121] See also Figure 4 ,The steps to evaluate the impact of users’ communication behavior in the game are:

[0122] Based on the user's in-game conversation behavior data, a deep learning model is used to extract keywords from the user's conversation information, analyze the user's in-game conversation information, and identify the user's conversation type, including normal game communication, normal life conversation, and abnormal conversation, to obtain the user conversation type recognition result;

[0123] Based on the user conversation type identification results, the frequency of each user conversation type is counted to obtain communication impact correlation data;

[0124] Based on the communication impact correlation data, the formula is:

[0125] ;

[0126] Calculate the user's communication impact index and obtain the user's communication impact assessment result;

[0127] in, The user's normal game communication frequency, The frequency of normal life conversations for users, is the user's abnormal conversation frequency, 、 and is the impact factor, The communication impact index of the user.

[0128] formula:

[0129] ;

[0130] Parameter details and how to obtain them:

[0131] The user's normal game communication frequency, The frequency of normal life conversations for users, The frequency of abnormal conversations among users. This is calculated by dividing the number of conversations of each type by the corresponding time period (e.g., the total time in a day). This reflects the user's activity in different conversation types.

[0132] 、 、 : The impact factor corresponding to each conversation type reflects the weight of different conversation types in evaluating user communication behavior. The factor is pre-set based on the social influence and importance of the conversation type, and is usually determined based on historical data or expert opinion.

[0133] Calculation example:

[0134] Assume the following data: In a day, a user has 100 normal game conversations, 150 normal life conversations, and 10 abnormal conversations. The time period is 24 hours. The conversation frequency is calculated as follows: ; ; ; Set the impact factor to , , .

[0135] calculate :

[0136]

[0137]

[0138]

[0139]

[0140] ;

[0141] The calculated result represents the user's overall communication influence index, which reflects the quality and magnitude of the user's social interactions within the game. A higher value indicates a user with greater communication influence within the game.

[0142] See also Figure 5 ,The steps to obtain the results of user social behavior analysis are:

[0143] Based on the user communication impact assessment results, the number of game friends of the user is extracted, and the average number of friends of all users is calculated to obtain social influence correlation data;

[0144] Based on social influence correlation data, the formula:

[0145] ;

[0146] Calculate the user's social influence index and obtain the user's social behavior analysis results;

[0147] in, is the average number of friends of game users, is the user's communication impact index, is the number of friends of the user, is the average activity of the user’s friends, The social influence index of the user.

[0148] formula:

[0149] ;

[0150] Parameter details and how to obtain them:

[0151] is the user's communication influence index, calculated through the previous steps.

[0152] The number of friends of the user is directly extracted by accessing the social network database of the game to count the total number of the user's current friends.

[0153] The average activity of the user's friends, obtained by aggregating the activity of all the user's friends and then taking the average.

[0154] It is the average number of friends of game users. The number of friends of all users in the game is queried in the database and the average value is calculated for standardized comparison.

[0155] Calculation example:

[0156] Set the user's communication influence index , the number of friends of this user , the average number of friends in the entire game , average activity of friends .

[0157] Calculate the ratio of friends:

[0158] ;

[0159] Calculate the logarithmic impact of friend activity:

[0160] 8;

[0161] calculate :

[0162] ;

[0163] The calculated result represents the user's social influence index, quantifying their social influence within the game. A higher value indicates that the user not only communicates frequently in-game but also has a large number of active friends, thus possessing greater influence within the gaming social network. This approach provides a comprehensive assessment of a user's in-game social behavior by combining the quality of communication, social breadth, and social depth.

[0164] See also Figure 6 ,The steps to evaluate the importance level of each user are:

[0165] Based on the user activity analysis results and the user social behavior analysis results, the user's activity index, stability index and social influence index are extracted, and the preset minimum importance level is extracted to obtain importance-related data;

[0166] Based on the importance correlation data, the formula is:

[0167] ;

[0168] Calculate the user's importance level and obtain the user's importance evaluation result;

[0169] in, is the preset minimum importance level, is the user's activity index, is the user's stability index, is the user's social influence index, 、 and is the weight coefficient, The importance level of the user.

[0170] formula:

[0171] ;

[0172] Parameter details and how to obtain them:

[0173] : The user's activity index, calculated in the previous step.

[0174] : The user's stability index, calculated in the previous step.

[0175] : The user's social influence index, calculated through the previous steps.

[0176] 、 、 : The weight coefficient represents the relative importance of each index in calculating the overall influence, and is usually set based on business needs and historical data.

[0177] : The preset minimum importance level, ensuring that the user level is not lower than this value to fairly reflect the basic influence of all users.

[0178] Calculation example:

[0179] Set a user's activity index , stability index , social influence index , the weight coefficients are set as , , , minimum importance level .

[0180] The calculation process is as follows:

[0181] Calculate the weighted sum of the indices:

[0182]

[0183]

[0184]

[0185] ;

[0186] Convert the values to logarithmic form to increase the nonlinear response, and calculate the integer part:

[0187] ;

[0188] ;

[0189] Minimum importance level limit:

[0190] ;

[0191] The calculation results show that although this user demonstrates a certain level of activity and social influence in the game, their final importance level is set to 2, taking into account the preset minimum level. This ensures that all users have at least basic recognition and importance, avoiding low ratings due to algorithmic sensitivity. This calculation method aims to balance the overall evaluation of user contributions with maintaining a certain level of baseline.

[0192] See also Figure 7 , the steps to obtain the user preference content recognition results are:

[0193] Based on the user importance assessment results, by counting the number of users with differentiated importance levels, analyzing the proportion of users with differentiated importance levels, and extracting behavioral data of users with differentiated importance levels, including the frequency and time of playing each game content, we can obtain game preference correlation data;

[0194] Based on the game preference association data, the formula is:

[0195] ;

[0196] Calculate the user's preference index for game content;

[0197] in, For the The proportion of users with different importance levels, For the The number of times users of different importance levels participate in game content, For the The total playing time of users with different levels of importance on game content, For the The value of the importance level, is the total number of importance levels, The user's preference index for game content;

[0198] Based on the user's preference index for game content, the game content is sorted according to the size of the preference index, the game content preferred by the user is identified, and the user's preferred content identification result is obtained.

[0199] formula:

[0200] ;

[0201] Parameter details and how to obtain them:

[0202] : No. The proportion of users of each importance level is obtained by analyzing the user database and calculating the proportion of users of each importance level to the total number of users.

[0203] : No. The number of times users of different importance levels participate in game content, that is, the number of times users participate in specific game content. This is usually calculated by tracking user game behavior logs.

[0204] : No. The total play time of users with different importance levels on game content, that is, the total play time of users on specific game content. This is also calculated by analyzing users' game behavior logs.

[0205] : No. The value of the importance level is calculated by the previous step.

[0206] : The total number of importance levels, as determined by the game or platform's user rating policy.

[0207] Calculation example:

[0208] There are three different importance levels ( ), the user percentage, participation times, play time and level values of each level are as follows: Level 1: , , Hour, Level 2: , , Hour, 2; Level 3: , , Hour, .

[0209] The calculation process is as follows:

[0210]

[0211]

[0212]

[0213]

[0214] ;

[0215] The calculation results show that the game content is very popular among users of different levels, especially among users with higher importance levels. This calculation method provides a deeper understanding of the popularity of game content among different user levels, helping game developers and marketing strategists optimize products and services for different user groups.

[0216] See also Figure 8 , the steps to obtain the game update impact analysis results are:

[0217] Based on the results of user preference content identification, game updates are implemented according to the user's preference content, and user activity index and stability index data before and after the update are extracted. The changes in user participation in the target game content, including changes in play frequency and time, are also extracted to obtain update effect correlation data;

[0218] Based on the update effect associated data, through the formula:

[0219] ;

[0220] Calculate the game update effect score;

[0221] in, Rate the game update performance, is the change in activity index, is the change in stability index, The change in frequency of game participation for target update content, The amount of game time change for the target update content, 、 and is the weight coefficient, is a symbolic function that returns The symbol, when When is a positive number, For the right, when When it is a negative number, is negative;

[0222] Based on the game update effect score, the game update effect is evaluated according to the positive or negative and size of the score, and it is determined whether the game update effect meets expectations, and the game update impact analysis results are obtained.

[0223] formula:

[0224] ;

[0225] Parameter details and how to obtain them:

[0226] : The change in activity index, obtained by comparing the average activity index of users before and after the game update, calculated as the index after the update minus the index before the update.

[0227] : Change in stability index. Similar to activity, this is obtained by comparing the average stability index of users before and after the update.

[0228] and : The change in frequency of game participation for target update content, The change in game time for the target updated content is obtained through log analysis by counting the differences in the average play time and average play frequency of all users before and after the update of specific game content.

[0229] : Sign function, used to determine the positive and negative signs of . This helps to correctly handle the impact of increases and decreases in frequency and time on the total score.

[0230] 、 、 : Weight coefficients corresponding to the influence of changes in activity, stability, and game engagement. These coefficients are determined based on business needs and historical data analysis to ensure that each factor has an appropriate impact on the score.

[0231] Calculation example:

[0232] The user's index before and after the game update is set as follows: Activity index change before and after the update , stability index changes before and after the update , the change in participation frequency of target game content , the change in play time of the target game content Hours. Set the weight coefficient to: , , .

[0233] Calculating game update effectiveness ratings :

[0234]

[0235]

[0236]

[0237]

[0238] ;

[0239] Calculation results , indicating that the game update had a negative impact, particularly a significant decrease in game participation frequency and time. This negative result suggests that the game update may not have achieved the expected effect of attracting users, or the updated content may not be in line with the user's preferences.

[0240] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A game behavior analysis system based on deep learning, characterized in that: The system comprises: The game activity analysis module counts users' daily play time based on their gaming behavior data, analyzes their average weekly online times, evaluates their activity, analyzes their activity data, evaluates their stability, and obtains user activity analysis results. The steps of evaluating the user's activity are: Based on the user's gaming behavior data, the user's playing time is counted, and the user's average weekly online times are analyzed to obtain activity-related data; Based on the activity-related data, the formula: ； Calculate the user's activity index; in, The user's daily play time, The total number of times the user has been online in the past week. and is the weight coefficient, is the user's activity index; Based on the activity index of the user, the activity of the user is evaluated according to the size of the activity index to obtain user activity information; The steps for obtaining the user activity analysis results are: Based on the user activity information, extracting user activity data within a period of time according to a preset time interval to obtain stability correlation information; Based on the stability association information, the formula: ； Calculate the user's stability index; in, is the user's stability index, is the total number of days in the statistical period, For the User activity index for the day, is the average activity index; Based on the user's stability index, the user's stability is evaluated according to the size of the stability index, and the user's activity index is integrated to obtain a user activity analysis result; The game social behavior analysis module uses a deep learning model to analyze user conversation behavior data in the game, identify user conversation types, and evaluate the impact of user communication behavior in the game. Combined with the number of friends a user has in the game, it evaluates the user's social influence in the game and obtains the user's social behavior analysis results. The steps for evaluating the impact of user communication behavior in the game are: Based on the user's in-game conversation behavior data, a deep learning model is used to extract keywords from the user's conversation information, analyze the user's in-game conversation information, and identify the user's conversation type, including normal game communication, normal life conversation, and abnormal conversation, to obtain the user conversation type recognition result; Based on the user conversation type identification results, the frequency of each user conversation type is counted to obtain communication impact correlation data; Based on the communication impact correlation data, the formula: ； Calculate the user's communication impact index and obtain the user's communication impact assessment result; in, The user's normal game communication frequency, The frequency of normal life conversations for users, is the user's abnormal conversation frequency, 、 and is the impact factor, Communication impact index for users; The steps for obtaining the user social behavior analysis results are: Based on the user communication impact evaluation results, extract the number of game friends of the user and calculate the average number of friends of all users to obtain social influence correlation data; Based on the social influence correlation data, the formula: ； Calculate the user's social influence index and obtain the user's social behavior analysis results; in, is the average number of friends of game users, is the user's communication impact index, is the number of friends of the user, is the average activity of the user’s friends, The social influence index of the user; The user preference identification module evaluates the importance level of each user based on the user's activity, stability, and social influence based on the user's activity and social behavior analysis results, extracts the user's behavior data, evaluates the overall user's preference for differentiated game content, identifies the user's preferred game content, and obtains a user preference content identification result; The steps of evaluating the importance level of each user are: Based on the user activity analysis results and the user social behavior analysis results, extract the user's activity index, stability index and social influence index, and extract the preset minimum importance level to obtain importance-related data; Based on the importance associated data, the formula: ； Calculate the user's importance level and obtain the user's importance evaluation result; in, is the preset minimum importance level, is the user's activity index, is the user's stability index, is the user's social influence index, 、 and is the weight coefficient, is the user's importance level; The steps for obtaining the user preference content recognition result are: Based on the user importance evaluation results, by counting the number of users with differentiated importance levels, analyzing the proportion of users with differentiated importance levels, and extracting behavioral data of users with differentiated importance levels, including the frequency and time of playing each type of game content, to obtain game preference association data; Based on the game preference association data, the formula: ； Calculate the user's preference index for game content; in, For the The proportion of users with different importance levels, For the The number of times users of different importance levels participate in game content, For the The total playing time of users with different levels of importance on game content, For the The value of the importance level, is the total number of importance levels, The user's preference index for game content; Based on the user's preference index for game content, sorting the game content according to the size of the preference index, identifying the game content preferred by the user, and obtaining a user preferred content identification result; The update impact analysis module extracts the changes in user activity and stability before and after the game update based on the user preference content identification results, and combines the changes in the frequency and time of users playing differentiated game content to evaluate the effect of the game update and obtain the game update impact analysis results; The steps for obtaining the game update impact analysis results are as follows: Based on the user preference content identification results, implement game updates according to the user's preference content, and extract user activity index and stability index data before and after the update, and extract changes in user participation in the target game content, including changes in play frequency and time, to obtain update effect correlation data; Based on the update effect associated data, by the formula: ； Calculate the game update effect score; in, Rate the game update performance, is the change in activity index, is the change in stability index, The change in frequency of game participation for target update content, The amount of game time change for the target update content, 、 and is the weight coefficient, is a symbolic function that returns The symbol, when When is a positive number, For the right, when When it is a negative number, is negative; Based on the game update effect score, the game update effect is evaluated according to the positive or negative and size of the score, and it is determined whether the game update effect meets expectations, thereby obtaining a game update impact analysis result.

Images