A game guidance adjustment method based on human-computer game data

Abstract

The present invention discloses a game guidance adjustment method based on human-computer game data, including judging the user's learning stage through the user's historical human-computer game data; identifying the user's game behavior patterns at different learning stages through the user's historical game behavior data; designing a move prompt strategy according to the user's learning stage and game behavior pattern; judging the complexity of the current game according to real-time game data, the user's learning stage and the user's game behavior pattern, and adjusting the move prompt frequency and the move prompt detail level; determining the effectiveness score of the move prompt strategy through the user's real-time move prompt effect data, and optimizing the move prompt strategy. The present invention provides an intelligent and personalized Go learning guidance method, which significantly improves the user's overall learning experience and effect through a real-time and dynamic move prompt strategy, and effectively solves the problems of insufficient personalization, rigid prompt strategy, lack of effective feedback, etc. in the prior art.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a game guidance adjustment method based on human-computer game data. Background Art

[0002] As a strategic chess game with a long history, Go is loved by chess players all over the world. However, in the process of learning and training Go, traditional teaching methods have many limitations. With the advancement of science and technology, especially the development of artificial intelligence technology, Go teaching methods are gradually developing in the direction of intelligence. However, the existing Go learning system still has many problems in terms of move prompt strategy and personalized guidance, which seriously affects the user's learning experience and efficiency. It is difficult for the existing Go learning system to accurately judge the user's learning stage. Go learning is a gradual and progressive process, and learners at different stages have significant differences in skills, strategy understanding and decision-making ability. Traditional Go teaching mostly adopts a one-size-fits-all approach and cannot provide accurate teaching content and strategic guidance for players of different levels. This lack of personalized teaching method often makes it difficult for beginners to master basic skills, while advanced players find it difficult to break through bottlenecks, and the overall learning efficiency is low. Users have various behavior patterns in Go games, and existing systems are difficult to flexibly identify and adapt. In the process of Go learning, users' behavior patterns will change with different learning stages, such as relying on prompts, independent thinking, quick decision-making, deep thinking, etc. These behavior patterns have different guidance needs at different learning stages. However, most existing Go learning systems lack the ability to identify user behavior patterns and cannot adjust the prompt strategy according to the user's actual situation, resulting in a lack of targeted guidance for users during the learning process and poor learning results. During the Go game, the complexity of the game is an important factor affecting user learning. Go games are varied, and especially in complex situations, users often find it difficult to make correct decisions. Most of the existing move prompt strategies are based on simple rules or preset strategies, which cannot evaluate and adapt to the complexity changes of the game in real time. This fixed prompt strategy is difficult to provide effective guidance when facing complex situations, resulting in users not getting the help they need at critical moments, affecting the improvement of their game level. More importantly, the existing system lacks an effective feedback mechanism to evaluate the effectiveness of the move prompt strategy. Go learning is a dynamic adjustment process, and the prompt strategy needs to be continuously optimized based on the user's real-time performance. However, most current systems do not have a perfect mechanism to collect and analyze users' real-time move prompt effect data, and cannot evaluate the effectiveness of the prompt strategy, resulting in the prompt strategy not being adjusted and optimized with the user's progress. This not only affects the user's learning efficiency, but may also lead to a decrease in the user's trust in the system, further affecting their learning enthusiasm. Therefore, the existing Go learning system has significant deficiencies in personalized learning experience, behavioral pattern recognition, dynamic adjustment of prompt strategies, real-time evaluation of game complexity, and optimization of prompt strategy effectiveness. The existence of these problems seriously restricts the user's learning effect and improvement of the game level. In order to solve these problems, it is necessary to develop a more intelligent and personalized Go learning guidance method, provide real-time and dynamic move prompt strategies, and improve the user's overall learning experience and effect. Summary of the invention

[0003] The present invention solves the problems existing in the above-mentioned prior art and provides a game guidance adjustment method based on human-computer game data, which mainly includes:

[0004] Based on the user's historical human-computer game data, a learning stage recognition model is constructed to determine the user's learning stage;

[0005] Through the user's historical game behavior data, a user game behavior pattern recognition model is constructed to identify the game behavior patterns of users at different learning stages;

[0006] Design a move prompt strategy based on the user's learning stage and game behavior pattern, and adjust the move prompt frequency and the level of detail based on the changes in the user's learning stage and game behavior pattern;

[0007] According to the real-time chess game data, the user's learning stage and the user's game behavior pattern, a chess game complexity evaluation model is constructed to determine the current chess game complexity, and the frequency and detail of the move prompts are adjusted based on the complexity of the real-time game.

[0008] Through the user's real-time move prompt effect data, determine the effectiveness score of the move prompt strategy and optimize the move prompt strategy.

[0009] Furthermore, the method of constructing a learning stage identification model through the user's historical human-computer game data to determine the user's learning stage includes:

[0010] Through historical game records, the user's historical human-computer game data is obtained, and the historical game data is marked with learning stages. The human-computer game data includes the outcome of each game, the thinking time for each move, the overall game duration, the human-computer difficulty, and the learning stages include beginner stage, intermediate stage and advanced stage; based on the historical human-computer game data obtained, data cleaning technology is used to remove missing values ​​and outliers, and the data is standardized to obtain the cleaned historical human-computer game data; based on the cleaned historical human-computer game data, a decision tree algorithm is used to train the model, a learning stage recognition model is constructed, and the user's learning stage is determined; real-time human-computer game data is obtained, and a learning stage recognition model is used to determine the user's learning stage.

[0011] Furthermore, the user's historical game behavior data is used to construct a user game behavior pattern recognition model to identify the game behavior patterns of the user at different learning stages, including:

[0012] Through historical game records, the user's historical game behavior data is obtained, and the historical game behavior data is marked with the learning stage. The game behavior data includes the average thinking time for each game and the frequency of using the move prompts for each game. The average thinking time for each game = total thinking time for each game / total number of steps, and the frequency of using the move prompts for each game = number of times the move prompts are used in each game / total number of steps; based on the historical game behavior data obtained, data cleaning technology is used to remove missing values ​​and outliers, and the historical game behavior data is standardized; based on the historical game behavior data after data cleaning, the decision tree algorithm is used for model training, and a user game behavior pattern recognition model is constructed to identify the game behavior patterns of users at different learning stages. The game behavior patterns include prompt dependence type, independent thinking type, quick decision-making type, and deep thinking type.

[0013] Furthermore, the step-drop prompt strategy is designed according to the user's learning stage and game behavior pattern, and the step-drop prompt frequency and the step-drop prompt detail level are adjusted based on the changes in the user's learning stage and game behavior pattern, including:

[0014] According to the user's learning stage and gaming behavior pattern, a move prompt strategy is designed, including setting the basic move prompt frequency and basic move prompt detail level for each learning stage, and adjusting the move prompt frequency and move prompt detail level based on the user's gaming behavior pattern; for users who rely on prompts, increase the move prompt frequency and move prompt detail level; for users with independent thinking, quick decision-making, and deep thinking gaming behavior patterns, reduce the prompt frequency and prompt detail level; based on the acquired real-time human-computer gaming data, use the learning stage recognition model and the user gaming behavior pattern recognition model to determine whether the user's learning stage and the user's gaming behavior pattern have changed; if the user's learning stage and gaming behavior pattern have changed, adjust the move prompt frequency and move prompt detail level.

[0015] Furthermore, the game complexity evaluation model is constructed according to the real-time game data, the user's learning stage and the user's game behavior pattern, the current game complexity is determined, and the frequency and detail of the move prompts are adjusted based on the complexity of the real-time game, including:

[0016] According to the user's real-time chess game records, real-time chess game data is obtained. The chess game data includes the number of moves, move information, chess piece distribution information, chess game innovation index, potential best moves, and the changing trend of the situation. The changing trend of the situation includes changes in the advantaged party and the territory control range of both parties; according to the chess game data, the user's learning stage and the user's behavioral game pattern, a chess game complexity evaluation model is used to build a chess game complexity evaluation model to judge the current chess game complexity, which includes simple, medium and complex; according to the real-time chess game data, the user's learning stage and the user's behavioral game pattern, the chess game complexity evaluation model is used to judge the current chess game complexity; if the current chess game complexity is complex, the move prompt frequency and the move prompt detail level are increased; according to the user's real-time game record, the chess game data is continuously obtained and updated, the chess game complexity is evaluated in real time using the chess game complexity evaluation model, and the move prompt frequency and the move prompt detail level are adjusted based on the real-time game game complexity.

[0017] It also includes constructing a chess game innovation index evaluation formula based on the current chess game move data and historical chess game move data to evaluate the degree of innovation of the current chess game.

[0018] The method of constructing a chess game innovation index evaluation formula based on the current chess game move data and the historical chess game move data to evaluate the innovation index of the current chess game specifically includes:

[0019] Through historical game records, historical move data of different learning stages and strategic chess games are obtained; through the cosine similarity calculation method, the matching degree between the current chess game move data and the move with the highest similarity in the historical move data is calculated; based on the historical move data and the current chess game move data, a chess game innovation index evaluation formula is constructed Calculate the game innovation index, where N is the number of moves made so far, and m is the number of moves made so far. i is the move at step i, h is the historical move dataset, P(m i , h) represents the matching degree between the i-th move and the most similar move in the historical move data, and M is the maximum possible matching degree, which is defined according to the chess type and historical data set.

[0020] Furthermore, the effectiveness score of the move prompt strategy is determined through the user's real-time move prompt effect data, and the move prompt strategy is optimized, including:

[0021] Through the user's game records, obtain the user's real-time move prompt effect data, and use the move prompt strategy effectiveness evaluation formula Determine the effectiveness score of the prompt strategy, where E is the effectiveness score of the prompt strategy, T r is the response time for each prompt, T t is the average thinking time of users, A ais the acceptance rate of the prompt, k is the smoothing factor for adjusting the acceptance rate, A0 is the equilibrium point of the acceptance rate, ranging from 0 to 1, Q p Score the user's performance after the prompt, Q e Score the user's incorrect performance after the prompt. The move prompt effect data includes the response time of each move prompt, the user's average thinking time, the move prompt acceptance rate, the performance score after the move prompt and the error performance score after the move prompt. The performance score after the move prompt and the error performance score after the move prompt are determined based on the degree of deviation from the optimal strategy. If the effectiveness score of the move prompt strategy is less than the preset effectiveness score threshold, identify the move prompt strategy part whose response time to the move prompt is greater than the preset time threshold, whose move prompt acceptance rate is lower than the preset acceptance rate threshold, and whose performance score after the move prompt is lower than the preset score threshold, and optimize the move prompt strategy parameters of the move prompt strategy part. Obtain the user's real-time game record, calculate the effectiveness score of the optimized move prompt strategy, and if the effectiveness score still does not exceed the preset effectiveness score threshold, further adjust and optimize the move prompt strategy until the effectiveness score of the move prompt strategy exceeds the preset effectiveness score threshold.

[0022] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0023] The present invention provides a game guidance adjustment method based on human-computer game data, which realizes personalized Go learning guidance by accurately identifying the user's learning stage and game behavior pattern. The system can provide a tailored drop prompt strategy according to the actual needs of the user, thereby significantly improving the learning efficiency and effect. By dynamically adjusting the drop prompt strategy, the system can timely adjust the prompt frequency and detail level when the user's learning stage and behavior pattern change, ensuring that the user obtains appropriate guidance when help is needed, and is not excessively intervened during autonomous learning, which not only improves the user's autonomous learning ability, but also stimulates the user's creativity and thinking ability. The present invention also dynamically adjusts the frequency and detail level of the drop prompt according to the complexity of the current chess game by real-time evaluation of the complexity of the chess game, so that the user can get more accurate and timely guidance when facing a complex situation, and improves his decision-making ability and game level. In addition, by continuously evaluating and optimizing the effectiveness of the drop prompt strategy, the system can continuously adjust and optimize the parameters of the prompt strategy according to the user's real-time performance, ensuring that the prompt strategy continues to improve with the user's progress. This not only improves the effectiveness and accuracy of the prompt strategy, but also enhances the user's learning experience and trust, and further stimulates the user's learning enthusiasm. The present invention provides an intelligent and personalized Go learning guidance method, which significantly improves the user's overall learning experience and effect through real-time and dynamic move prompt strategies, effectively solves the problems of insufficient personalization, rigid prompt strategies, lack of effective feedback, etc. in the prior art, and provides strong support for the intelligent development of Go learning. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A flowchart of a game guidance adjustment method based on human-computer game data of the present invention;

[0025] Figure 2 A schematic diagram of a game guidance adjustment method based on human-computer game data of the present invention;

[0026] Figure 3 It is another schematic diagram of a game guidance adjustment method based on human-computer game data of the present invention;

[0027] Figure 4 A schematic diagram of an interface of associated equipment of a game guidance adjustment method based on human-computer game data of the present invention;

[0028] Figure 5 This is another schematic diagram of an associated device interface of a game guidance adjustment method based on human-computer game data of the present invention. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0030] like Figure 4-5 The present invention relates to an electronic ink screen puzzle chess learning treasure that integrates multiple functions, aiming to provide users with a full range of learning and entertainment experience. It is not only an e-book reader, but also an intelligent device that integrates chess learning, puzzle games, poetry and MP3 playback and other multi-functions. The device adopts advanced electronic ink screen technology to reduce blue light damage, provide eye-protecting reading experience, protect eyesight, and is suitable for long-term use. In terms of chess learning, the device contains learning sections such as Go, Gobang and Othello. The Go learning section is powerful, including AI battles, chess strength tests, introductory training, level-breaking and life and death exercises, etc., providing users with comprehensive training from beginners to advanced chess players. This electronic ink screen puzzle chess learning treasure is not only a learning tool, but also an intelligent device that comprehensively improves users' intelligence, cultural literacy and entertainment experience, and is suitable for users of all ages.

[0031] like Figure 1-3 In this embodiment, a game guidance adjustment method based on human-computer game data may specifically include:

[0032] Step S101, constructing a learning stage recognition model based on the user's historical human-computer game data to determine the user's learning stage.

[0033] Through historical game records, the user's historical human-computer game data is obtained, and the historical game data is labeled with the learning stage. The human-computer game data includes the outcome of each game, the thinking time for each move, the overall game duration, and the difficulty of the human-computer game. The learning stage includes the beginner stage, the intermediate stage, and the advanced stage. Based on the historical human-computer game data obtained, data cleaning technology is used to remove missing values ​​and outliers, and the data is standardized to obtain the cleaned historical human-computer game data. Based on the cleaned historical human-computer game data, the decision tree algorithm is used for model training, and a learning stage identification model is constructed to determine the user's learning stage. Real-time human-computer game data is obtained, and the learning stage identification model is used to determine the user's learning stage.

[0034] Exemplarily, through the historical game records, the user's historical human-computer game data is obtained, including the results of 20 games, the average thinking time for each move, the overall game duration of each game, and the human-computer difficulty ranging from 1 to 5. These data are labeled to determine the learning stage of the user, which includes the beginner stage, the intermediate stage, and the advanced stage. Use data cleaning technology to remove missing values ​​and outliers, such as data with a game time of more than 1 hour or less than 5 minutes in a certain game is cleared, and the remaining data is standardized to obtain the cleaned historical human-computer game data. Using the cleaned data, the model is trained through the decision tree algorithm to construct a learning stage recognition model. Obtain the user's real-time human-computer game data, such as the average thinking time for the latest game is 14 seconds, the total game duration is 24 minutes, and the human-computer difficulty is 3. Use the previously trained learning stage recognition model to determine that the user is currently in the intermediate stage.

[0035] Step S102, constructing a user game behavior pattern recognition model based on the user's historical game behavior data to identify the game behavior patterns of the user at different learning stages.

[0036] Through historical game records, the user's historical game behavior data is obtained, and the historical game behavior data is marked with the learning stage. The game behavior data includes the average thinking time for each game and the frequency of using the move prompts for each game. The average thinking time for each game = the total thinking time for each game / the total number of steps, and the frequency of using the move prompts for each game = the number of times the move prompts are used in each game / the total number of steps. Based on the historical game behavior data obtained, data cleaning technology is used to remove missing values ​​and outliers, and the historical game behavior data is standardized. Based on the historical game behavior data after data cleaning, the decision tree algorithm is used for model training, and a user game behavior pattern recognition model is constructed to identify the game behavior patterns of users at different learning stages. Game behavior patterns include prompt-dependent, independent thinking, quick decision-making, and deep thinking.

[0037] Exemplarily, through the historical game records, the historical game behavior data of user C is obtained, and these data are labeled in the learning stage. The game behavior data includes the average thinking time for each game and the frequency of using the prompt for each game, where the average thinking time for each game is calculated as the total thinking time for each game divided by the total number of steps, and the frequency of using the prompt for each game is calculated as the number of times the prompt for each game is used divided by the total number of steps. Therefore, in the first 30 games of user C, the average thinking time for each game is 20 seconds, the total number of steps is 50, the total thinking time is 1000 seconds, and the number of times the prompt for the game is used is 10 times, so the average thinking time for each game is 20 seconds, and the frequency of using the prompt for the game is 0.2. In the next 20 games, the average thinking time for each game is 15 seconds, the total number of steps is 60, the total thinking time is 900 seconds, and the number of times the prompt for the game is used is 5 times, so the average thinking time for each game is 15 seconds, and the frequency of using the prompt for the game is 0.083. In the next 10 games, the average thinking time for each game was 10 seconds, the total number of steps was 70, the total thinking time was 700 seconds, and the number of times the prompt was used was 2 times. Therefore, the average thinking time for each game was 10 seconds, and the frequency of using the prompt was 0.028. Based on the historical game behavior data obtained, data cleaning technology was used to remove missing values ​​and outliers, and the data was standardized to ensure data consistency and comparability. Based on the cleaned data, the decision tree algorithm was used for model training to build a user game behavior pattern recognition model, and the game behavior patterns of user C at different learning stages were identified. The first 30 games were identified as dependent prompts, the middle 20 games were identified as independent thinking, and the last 10 games were identified as quick decision-making.

[0038] Step S103, designing a move prompt strategy according to the user's learning stage and game behavior pattern, and adjusting the move prompt frequency and the move prompt detail level based on changes in the user's learning stage and game behavior pattern.

[0039] Design a move prompt strategy based on the user's learning stage and game behavior pattern, including setting the basic move prompt frequency and basic move prompt detail level for each learning stage, and adjust the move prompt frequency and move prompt detail level based on the user's game behavior pattern. For users who rely on prompts, increase the move prompt frequency and move prompt detail level. For users with independent thinking, quick decision-making, and deep thinking game behavior patterns, reduce the prompt frequency and prompt detail level. Based on the real-time human-computer game data obtained, use the learning stage recognition model and the user game behavior pattern recognition model to determine whether the user's learning stage and the user's game behavior pattern have changed. If the user's learning stage and game behavior pattern have changed, adjust the move prompt frequency and move prompt detail level.

[0040] Exemplarily, according to the user's learning stage and game behavior pattern, a move prompt strategy is designed, including setting the basic move prompt frequency and basic move prompt detail level for each learning stage, and adjusting the move prompt frequency and move prompt detail level based on the user's game behavior pattern. If user F is in the beginner stage, the basic move prompt frequency is to give a prompt every time a complex situation is encountered, and the prompt detail level is to explain in detail the advantages and disadvantages and potential consequences of each optional move. User F is identified as a prompt-dependent type in the beginner stage. Therefore, during the game, when the system detects that the user is in a complex situation at steps 5 and 15, the prompt frequency and detail level are increased. The system not only explains the best move in detail, but also explains why other choices are not good. For example, at step 5, the system prompts: "Choosing point A can expand your territory and prevent opponents from forming forces in the area, while choosing point B, although it seems good, is easily blocked by opponents." When F's learning stage progresses to the intermediate stage, the frequency of basic move prompts is adjusted to only give prompts when the user makes an obvious wrong decision, and the prompt level is to briefly prompt the main strategy. For example, at steps 25 and 35, the system only points out the best move and briefly explains the strategy, such as choosing point C will help consolidate your defense. If user F's game behavior pattern is identified as independent thinking, the actual move prompt frequency is reduced, and prompts are only provided when the user has not made a move for a long time or has made consecutive wrong decisions. For example, at steps 30 and 40, when the system detects that the user has been thinking for too long or has made an obvious wrong decision, it will give a brief prompt: "The best choice is point D." When user F plays a game, real-time human-computer game data is obtained, and the learning stage recognition model is used to determine that the user's learning stage has changed to the advanced stage. Real-time game behavior data is obtained, and the user game behavior pattern recognition model is used to identify that the user's game behavior pattern has changed to a quick decision-making type. Based on this change, the move prompt strategy is adjusted to provide brief suggestions only in extremely critical situations or when the user has not made a decision for a long time, and the prompt details are reduced to only point out the best move. For example, at step 45, when the system detects that the situation is complex and the user has not made a decision for a long time, a brief prompt is provided: "The best choice is point E." Through these adjustments, the system can dynamically adapt to the changes in user F's learning stage and game behavior pattern, provide personalized move prompt strategies, and thus optimize the learning effect.

[0041] Step S104, constructing a chess game complexity evaluation model based on real-time chess game data, the user's learning stage and the user's game behavior pattern, judging the current chess game complexity, and adjusting the move prompt frequency and move prompt detail based on the real-time game game complexity.

[0042] According to the user's real-time game records, real-time chess game data is obtained. The chess game data includes the number of moves, information about the moves, information about the distribution of chess pieces, the game innovation index, the potential best moves, and the trend of the situation. The trend of the situation includes the changes in the advantage side and the territory control range of both sides. According to the chess game data, the user's learning stage and the user's behavioral game mode, a chess game complexity evaluation model is used to build a chess game complexity evaluation model to determine the current chess game complexity. The complexity includes simple, medium and complex. According to the real-time chess game data, the user's learning stage and the user's behavioral game mode, the chess game complexity evaluation model is used to determine the current chess game complexity. If the current chess game complexity is complex, the frequency of move prompts and the level of detail of move prompts are increased. According to the user's real-time game records, the chess game data is continuously obtained and updated, and the chess game complexity is evaluated in real time using the chess game complexity evaluation model, and the frequency of move prompts and the level of detail of move prompts are adjusted based on the complexity of the real-time game.

[0043] Exemplarily, according to the user's real-time game record, real-time chess game data is obtained. The chess game data includes the current number of moves is 30, the move information shows that the latest move is at the D4 position, the chess piece distribution information shows that the black chess controls the central area, the white chess has an advantage in the corner area, the chess game innovation index is 0.8, the potential best move is E5, and the trend of situation change shows that the black chess gradually gains an advantage. The territory control range of both sides is 55% for black and 45% for white. According to these chess game data, the user is currently in the intermediate stage, and the user's behavior game mode is independent thinking, the chess game complexity evaluation model constructed by the recurrent neural network is used for analysis, and the current chess game complexity is judged to be complex. Therefore, the system increases the frequency of move prompts and the level of detail of the move prompts. In the analysis after each user makes a move, the system prompts the user of the current situation changes and recommends the best strategy, such as: "The current situation is complex, it is recommended to place a move at E5 to maintain control of the central area, while preventing the opponent from posing a threat at D5." As the user continues to play, the real-time game record is updated, the number of moves increases to 40, the latest move is at E5, the black chess control range expands to 60%, and the white chess reduces to 40%. The trend of the situation still shows that the black chess maintains the advantage, but the chess game innovation index drops to 0.5. The chess game complexity evaluation model is used to judge again that the current chess game complexity is still complex. The system continues to provide detailed move suggestions and prompts users to consider defending the corner areas to consolidate the current advantage.

[0044] Among them, according to the current chess game move data and historical chess game move data, a chess game innovation index evaluation formula is constructed to evaluate the innovation index of the current chess game.

[0045] Through historical game records, historical move data of different learning stages and strategic chess games are obtained. Through the cosine similarity calculation method, the matching degree between the current chess game move data and the move with the highest similarity in the historical move data is calculated. Based on the historical move data and the current chess game move data, a chess game innovation index evaluation formula is constructed. Calculate the game innovation index, where N is the number of moves made so far, and m is the number of moves made so far. i is the move at step i, h is the historical move dataset, P(m i , h) represents the matching degree between the i-th move and the most similar move in the historical move data, and M is the maximum possible matching degree, which is defined according to the chess type and historical data set.

[0046] For example, historical chess game records are used to obtain historical chess move data at different learning stages and strategy games. The cosine similarity calculation method is used to calculate the matching degree between the current chess move data and the move with the highest similarity in the historical move data. If the current chess game has been played for 20 moves, the move m of the current 15th move is 15 is D5, and the most similar move m' is found in the historical move data set as E5. The calculated matching degree P(m 15 ,h) is 0.75, and the possible maximum matching degree M is 1.0. Based on the historical move data and the current game move data, the following game innovation index evaluation formula is used Calculate the innovation index of the chess game, where N is 20. The matching degrees of each move calculated and the most similar moves in the historical move data are: 0.9, 0.85, 0.78, 0.95, 0.8, 0.75, 0.88, 0.79, 0.92, 0.84, 0.81, 0.77, 0.85, 0.89, 0.75, 0.9, 0.88, 0.87, 0.86, 0.91. Substitute these matching degrees into the formula to calculate the innovation index of the current chess game I n =0.1475.

[0047] Step S105, determining the effectiveness score of the move prompt strategy through the user's real-time move prompt effect data, and optimizing the move prompt strategy.

[0048] Through the user's game records, obtain the user's real-time move prompt effect data, and use the move prompt strategy effectiveness evaluation formula Determine the effectiveness score of the prompt strategy, where E is the effectiveness score of the prompt strategy, T r is the response time for each prompt, T t is the average thinking time of users, A a is the acceptance rate of the prompt, k is the smoothing factor for adjusting the acceptance rate, A0 is the equilibrium point of the acceptance rate, ranging from 0 to 1, Q pScore the user's performance after the prompt, Q e Score the user's incorrect performance after the prompt. The move prompt effect data includes the response time of each move prompt, the user's average thinking time, the move prompt acceptance rate, the performance score after the move prompt, and the error performance score after the move prompt. The performance score after the move prompt and the error performance score after the move prompt are determined based on the degree of deviation from the optimal strategy. If the effectiveness score of the move prompt strategy is less than the preset effectiveness score threshold, identify the move prompt strategy part whose response time to the move prompt is greater than the preset time threshold, the move prompt acceptance rate is lower than the preset acceptance rate threshold, and the performance score after the move prompt is lower than the preset score threshold, and optimize the move prompt strategy parameters of the move prompt strategy part. Obtain the user's real-time game record, calculate the effectiveness score of the optimized move prompt strategy, and if the effectiveness score still does not exceed the preset effectiveness score threshold, further adjust and optimize the move prompt strategy until the effectiveness score of the move prompt strategy exceeds the preset effectiveness score threshold.

[0049] For example, through the game record of the user, the real-time prompt effect data of user G is obtained, including the response time T of each prompt r is 5 seconds, and the average thinking time of users is T t The acceptance rate of the prompt is A for 20 seconds. a is 0.6, and the user's performance score after the prompt is Q p The score is 80, and the user's error performance score after the prompt is Q e =20 points. Use the formula to evaluate the effectiveness of the move hint strategy The effectiveness score E of the move prompt strategy is obtained. If the smoothing factor k of the adjusted acceptance rate is 0.5 and the equilibrium point A0 of the acceptance rate is 0.7, E=0.187 is obtained, which does not reach the preset effectiveness score threshold of 0.3. Therefore, according to the effectiveness evaluation of the move prompt strategy, existing problems are identified, such as the response time of 5 seconds is greater than the preset time threshold of 4 seconds, the move prompt acceptance rate of 0.6 is lower than the preset acceptance rate threshold of 0.7, and the performance score of 80 points after the move prompt is lower than the preset score threshold of 90 points. In response to these problems, the parameters of the move prompt strategy are optimized, such as shortening the response time, improving the accuracy of the prompt and user acceptance, and improving the performance evaluation method after the prompt. After optimization, the user's real-time game record is obtained, and the effectiveness score E=0.27 of the optimized move prompt strategy is recalculated. The effectiveness score after optimization still does not reach the preset effectiveness score threshold of 0.3. Therefore, we will continue to further adjust and optimize the move prompt strategy, such as further shortening the response time to 3 seconds, improving the prompt accuracy so that the acceptance rate reaches 0.8, and improving the performance score evaluation method after the prompt, so that the effectiveness score can eventually reach the expected standard.

[0050] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the concept of the present application. For example, the above features are replaced with the technical features with similar functions disclosed in the present application (but not limited to) to form a technical solution.

Claims

1. A game guidance adjustment method based on human-computer game data, characterized in that: The method comprises: Based on the user's historical human-computer game data, a learning stage recognition model is constructed to determine the user's learning stage; Through the user's historical game behavior data, a user game behavior pattern recognition model is constructed to identify the game behavior patterns of users at different learning stages; Design a move prompt strategy based on the user's learning stage and game behavior pattern, and adjust the move prompt frequency and the level of detail based on the changes in the user's learning stage and game behavior pattern; According to the real-time chess game data, the user's learning stage and the user's game behavior pattern, a chess game complexity evaluation model is constructed to determine the current chess game complexity, and the frequency and detail of the move prompts are adjusted based on the complexity of the real-time game. Determine the effectiveness score of the move prompt strategy through the user's real-time move prompt effect data, and optimize the move prompt strategy; The method of constructing a chess game complexity evaluation model based on real-time chess game data, the user's learning stage and the user's game behavior pattern, judging the complexity of the current chess game, and adjusting the frequency and detail of the move prompts based on the complexity of the real-time game, includes: Based on the current chess game move data and historical move data, a chess game innovation index evaluation formula is constructed to evaluate the innovation index of the current chess game; based on the user's real-time game records, real-time chess game data is obtained, and the chess game data includes the number of moves, move information, chess piece distribution information, chess game innovation index, potential best moves, and the trend of changes in the situation. The trend of changes in the situation includes changes in the advantaged party and the territory control range of both parties; based on the chess game data, the user's learning stage and the user's behavioral game mode, a chess game complexity evaluation model is used to construct a chess game complexity evaluation model to determine the complexity of the current chess game, and the complexity includes simple, medium and complex; based on the real-time chess game data, the user's learning stage and the user's behavioral game mode, the chess game complexity evaluation model is used to determine the complexity of the current chess game; if the current chess game complexity is complex, the move prompt frequency and the move prompt detail level are increased; based on the user's real-time game records, the chess game data is continuously obtained and updated, the chess game complexity is evaluated in real time using the chess game complexity evaluation model, and the move prompt frequency and the move prompt detail level are adjusted based on the complexity of the real-time game.

2. The method according to claim 1, wherein: The method constructs a learning stage identification model based on the user's historical human-computer game data to determine the user's learning stage, which is characterized by: Through historical game records, the user's historical human-computer game data is obtained, and the historical game data is marked with learning stages. The human-computer game data includes the outcome of each game, the thinking time for each move, the overall game duration, the human-computer difficulty, and the learning stages include beginner stage, intermediate stage and advanced stage; based on the historical human-computer game data obtained, data cleaning technology is used to remove missing values ​​and outliers, and the data is standardized to obtain the cleaned historical human-computer game data; based on the cleaned historical human-computer game data, a decision tree algorithm is used to train the model, a learning stage recognition model is constructed, and the user's learning stage is determined; real-time human-computer game data is obtained, and a learning stage recognition model is used to determine the user's learning stage.

3. The method according to claim 1, wherein: The method constructs a user game behavior pattern recognition model through the user's historical game behavior data to identify the game behavior patterns of users at different learning stages, and is characterized by: Through historical game records, the user's historical game behavior data is obtained, and the historical game behavior data is marked with the learning stage. The game behavior data includes the average thinking time for each game and the frequency of using the move prompts for each game. The average thinking time for each game = the total thinking time for each game / the total number of steps, and the frequency of using the move prompts for each game = the number of times the move prompts are used in each game / the total number of steps; based on the historical game behavior data obtained, data cleaning technology is used to remove missing values ​​and outliers, and the historical game behavior data is standardized; based on the historical game behavior data after data cleaning, the decision tree algorithm is used for model training, and a user game behavior pattern recognition model is constructed to identify the game behavior patterns of users in different learning stages. The game behavior patterns include prompt dependence type, independent thinking type, quick decision-making type, and deep thinking type.

4. The method according to claim 3, wherein: The method designs a move prompt strategy according to the user's learning stage and game behavior pattern, and adjusts the move prompt frequency and the move prompt detail level based on the user's learning stage and game behavior pattern changes, and is characterized by: According to the user's learning stage and gaming behavior pattern, a move prompt strategy is designed, including setting the basic move prompt frequency and basic move prompt detail level for each learning stage, and adjusting the move prompt frequency and move prompt detail level based on the user's gaming behavior pattern; for users who rely on prompts, increase the move prompt frequency and move prompt detail level; for users with independent thinking, quick decision-making, and deep thinking gaming behavior patterns, reduce the prompt frequency and prompt detail level; based on the acquired real-time human-computer gaming data, use the learning stage recognition model and the user gaming behavior pattern recognition model to determine whether the user's learning stage and the user's gaming behavior pattern have changed; if the user's learning stage and gaming behavior pattern have changed, adjust the move prompt frequency and move prompt detail level.

5. The method according to claim 1, wherein: The method constructs a chess game innovation index evaluation formula based on the current chess game move data and the historical chess game move data to evaluate the innovation index of the current chess game, which is characterized by: Through historical game records, historical move data of different learning stages and strategic chess games are obtained; through the cosine similarity calculation method, the matching degree between the current chess game move data and the move with the highest similarity in the historical move data is calculated; based on the historical move data and the current chess game move data, a chess game innovation index evaluation formula is constructed Calculate the game innovation index, where N is the number of moves made so far, and m is the number of moves made so far. i is the move at step i, h is the historical move dataset, P(m i , h) represents the matching degree between the i-th move and the most similar move in the historical move data, and M is the maximum possible matching degree, which is defined according to the chess type and historical data set.

6. The method according to claim 1, wherein: The effectiveness score of the move prompt strategy is determined through the user's real-time move prompt effect data, and the move prompt strategy is optimized, characterized in that: Through the user's game records, obtain the user's real-time move prompt effect data, and use the move prompt strategy effectiveness evaluation formula , determine the effectiveness score of the prompt strategy, where E is the effectiveness score of the prompt strategy, T r is the response time for each prompt, T t is the average thinking time of users, is the acceptance rate of the prompt, k is the smoothing factor for adjusting the acceptance rate, A0 is the equilibrium point of the acceptance rate, ranging from 0 to 1, Q p Score the user's performance after the prompt, Q e Score the user's incorrect performance after the prompt. The move prompt effect data includes the response time of each move prompt, the user's average thinking time, the move prompt acceptance rate, the performance score after the move prompt and the error performance score after the move prompt. The performance score after the move prompt and the error performance score after the move prompt are determined based on the degree of deviation from the optimal strategy. If the effectiveness score of the move prompt strategy is less than the preset effectiveness score threshold, identify the move prompt strategy part whose response time to the move prompt is greater than the preset time threshold, whose move prompt acceptance rate is lower than the preset acceptance rate threshold, and whose performance score after the move prompt is lower than the preset score threshold, and optimize the move prompt strategy parameters of the move prompt strategy part. Obtain the user's real-time game record, calculate the effectiveness score of the optimized move prompt strategy, and if the effectiveness score still does not exceed the preset effectiveness score threshold, further adjust and optimize the move prompt strategy until the effectiveness score of the move prompt strategy exceeds the preset effectiveness score threshold.

Images