A multi-person interaction system and method based on VR environment

Abstract

The present invention discloses a multiplayer interaction system and method based on a VR environment, and relates to the field of virtual reality technology, including: forming a player database; based on the player database, analyzing the preferred behavior style index of each player according to the action events in the player data; obtaining the game scene corresponding to the real-time progress in the linear process of the game, analyzing the task type in the game scene, and evaluating the difficulty index of the game scene; performing correlation matching screening on the preferred behavior style index of each player according to the difficulty index of the game scene, and generating the optimal player combination of the game scene according to the progress of the game process; generating a guide path scene according to the optimal player combination of the task of the progress of the game process, and completing the game interaction matching according to the current game space coordinate position of the player combination. The advantages of the present invention are: improving the interaction frequency in the VR game scene and improving the game immersion.

Description

Technical Field

[0001] The present invention relates to the field of virtual reality technology, and in particular to a multi-person interaction system and method based on a VR environment. Background Art

[0002] Multi-person interaction in a VR environment means that in an environment built with virtual reality (VR) technology, multiple users can interact, communicate and collaborate in real time to jointly experience the scenes and activities of the virtual world.

[0003] As existing VR games do not have perfect guidance tasks and matching mechanisms, users' actual experience is poor, VR game challenge scenes are stuck and difficult to get started, which prevents the immersive experience of VR from being fully utilized. Summary of the invention

[0004] In order to solve the above technical problems, a multiplayer interaction system and method based on a VR environment are provided. This technical solution solves the above-mentioned problem that due to the fact that the existing VR games have not yet perfected the guidance tasks and matching mechanisms, the actual user experience is poor, resulting in the VR game's breakthrough scenes being stuck and difficult to get started, and the VR immersive experience cannot be fully utilized.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] A multi-person interaction method based on a VR environment, comprising:

[0007] Based on the test levels, collect the player's playing data in the VR game and build a player database;

[0008] Based on the player database, analyze the preferred behavior style indicators of each player according to the action events in the player data;

[0009] Based on the game design documents, obtain the game scenes corresponding to the real-time progress in the linear process of the game, analyze the task types in the game scenes, and evaluate the difficulty indicators of the game scenes;

[0010] According to the difficulty index of the game scene, the preferred behavior style index of each player is screened for correlation matching to generate the optimal player combination for the game scene of the game process progress;

[0011] According to the optimal personnel combination for the task in the progress of the game process, and based on the current game space coordinate position of the personnel combination, a guidance path scene is generated to complete the game interaction matching.

[0012] It is understandable that since the differences between VR game content and desktop and mobile terminals are highlighted in the interaction and immersive experience, if mask loading is performed after player matching in the real-time game scene, there will be a gameplay vacuum period, resulting in a decrease in the VR immersive experience. Therefore, the matching combination is searched through the player's preferred ability and the difficulty of the task type, and a guidance path is generated to solve the problem of scene separation during matching loading.

[0013] Preferably, based on the player database, according to the action events in the player data, analyzing the preferred behavior style index of each player specifically includes:

[0014] Based on the action events in the player data, the action events are initially divided according to multi-dimensional directions to form a multi-dimensional data matrix of the players. A ; The multi-dimensional directions include: task exploration events, skill triggering events, and game triggering events;

[0015] ;

[0016] in, For the Player's Dimensional behavioral data parameters, is the total number of players, is the total number of multi-dimensional directional data;

[0017] Based on the game background database, the player's style and behavior feature data and label data are obtained, and the player's style and behavior feature data is used as a multi-dimensional original sample data set of the game players;

[0018] Based on random forest, the multi-dimensional original sample data set of game players is used as input, and the multi-dimensional original samples contained in the label data are used to train the corresponding game multi-dimensional behavior style recognition decision tree. The minimization of the label data classification error function of the multi-dimensional original sample data division result is used as the training goal, and the combination is a game multi-dimensional behavior style recognition model;

[0019] Substitute each element in the multi-dimensional data matrix of the players into the multi-dimensional game behavior style recognition model, use the maximum return Gini index of the multi-dimensional data of the players for the branch and leaf nodes of the multi-dimensional game behavior style recognition model as the segmentation condition, and generate the player's preferred behavior style index;

[0020] The preferred behavior style index of the players is generated as follows:

[0021] ;

[0022] In the formula, is the preferred behavior style indicator of the players, Identify the model function for the multi-dimensional behavior style of the game. A It is a multi-dimensional data matrix of players. For the Player's Dimensional behavioral data parameters, Divide the elements in the multi-dimensional data matrix of the players into k The proportion of data samples in each behavioral style category.

[0023] Preferably, based on the game design document, the game scene corresponding to the real-time progress in the linear process of the game is obtained, the task type in the game scene is analyzed, and the difficulty index of the game scene is evaluated, which specifically includes:

[0024] Based on the game design document, traverse the line process in the stable game design, uniquely associate and bind the scene identifier with the corresponding progress information, and create a game scene identifier progress information mapping function;

[0025] Based on the game backend database, the player's playing feedback data in each game scene is obtained, and according to the scene identifier, the player's playing progress actual operation feedback data array is obtained; the player's playing feedback data includes: level passing time, level failure number, and level failure frequency;

[0026] Using the actual operation feedback data array of the player's game progress, the scene identifier is used as the unique key to search, update the game scene identifier progress information mapping function, and establish the game scene identifier associated task type progress information mapping matrix B ;

[0027] ;

[0028] in, For the The identifier of the game scene corresponds to the The progress information of the associated task type, is the total number of scene identifiers, The total number of progress information of the associated task type;

[0029] Normalize each element in the game scene identifier associated task type progress information mapping matrix to obtain the game scene identifier associated task type progress information mapping normalized matrix ;

[0030] ;

[0031] in, For the The identifier of the game scene corresponds to the Normalized value of progress information of associated task types;

[0032] Based on the entropy weight method, a weight is assigned to each associated task type according to the overall proportion of the normalized value of the progress information of the task type associated with the game scene identifier, and the weight of the task type associated with the game scene identifier is obtained;

[0033] Based on the game scene identifier-associated task type progress information mapping, each element in the normalized matrix is ​​associated with the task type weight associated with the game scene identifier, and the task type difficulty index in the game scene is calculated;

[0034] The calculation of the difficulty index of the task type in the game scene is specifically:

[0035] ;

[0036] In the formula, The first The difficulty index of each game scene, For the The weight of each associated task type.

[0037] Preferably, the preferred behavior style index of each player is selected for correlation matching according to the difficulty index of the game scene, and the optimal player combination of the game scene for generating the game process progress specifically includes:

[0038] Based on cosine similarity, the matching degree between the preferred behavior style index of each player and the difficulty index of the game scene is calculated to construct the game scene matching matrix C of the players;

[0039] ;

[0040] In the formula, For the The preferred behavior style index of each player is related to the The matching value between the difficulty indicators of the game scenes;

[0041] Based on the game scene matching matrix of the game players, the lower limit of the number of team members in each game scene is used as a restriction condition;

[0042] The objective function is to search for the player matching value that maximizes the current game scenario;

[0043] Construct a model for screening game personnel in the game scene;

[0044] Based on the player screening model in the game scene, each element in the player's game scene matching matrix is ​​used as input, and the optimal player combination of the game scene in the game process progress is used as output;

[0045] The optimal player combination for the game scene of the game process is as follows:

[0046] ;

[0047] In the formula, is a binary decision variable (0,1) when it is 1, Indicates the player Assigned to game scene middle, Indicates the player Not assigned to a game scene middle, It is the lower limit of the number of team members in the game scene. Assign unique constraints to people, Assign unique player count constraints to game scenarios.

[0048] Furthermore, a multi-person interaction system based on a VR environment is proposed, which is used to implement the multi-person interaction method based on a VR environment as described above, comprising:

[0049] Test database: Based on the test levels, collect the player play data in the VR game and build a player database;

[0050] A player style evaluation module, the style evaluation module is electrically connected to the test database and is based on the player database, and analyzes the preferred behavior style indicators of each player according to the action events in the player data;

[0051] The game difficulty assessment module is used to obtain the game scene corresponding to the real-time progress in the linear process of the game based on the game design document, analyze the task type in the game scene, and evaluate the difficulty index of the game scene;

[0052] A personnel screening module, which is electrically connected to the personnel style assessment module and the game difficulty assessment module. The personnel screening module is used to perform correlation matching screening based on the preferred behavior style indicators of each player according to the difficulty indicators of the game scene, and generate the optimal personnel combination for the game scene of the game process progress;

[0053] The guidance module is electrically connected to the personnel screening module. The guidance module is used to generate a guidance path scene based on the optimal personnel combination for the task according to the progress of the game process and the current game space coordinate position of the personnel combination to complete the game interaction matching.

[0054] Optionally, the personnel style assessment module includes:

[0055] The multi-dimensional matrix unit performs initial division of the action events according to the multi-dimensional directions based on the action events in the player data, and forms a multi-dimensional data matrix of the players;

[0056] The feedback big data unit obtains the player's style behavior feature data and label data based on the game background database, and uses the player's style behavior feature data as a multi-dimensional original sample data set of the game player;

[0057] The first model building unit, based on random forest, takes the multi-dimensional original sample data set of game players as input, takes the multi-dimensional original samples contained in the label data, trains the corresponding game multi-dimensional behavior style recognition decision tree, takes the minimization of the label data classification error function of the multi-dimensional original sample data division result as the training goal, and combines them into a game multi-dimensional behavior style recognition model;

[0058] The first model analysis unit substitutes each element in the multi-dimensional data matrix of the players into the multi-dimensional behavior style recognition model of the game, and uses the maximum return Gini index of the multi-dimensional data of the players for the branch and leaf nodes of the multi-dimensional behavior style recognition model of the game as a segmentation condition to generate the player's preferred behavior style index.

[0059] Optionally, the game difficulty assessment module includes:

[0060] A mapping function unit, based on the game design document, traverses the line flow in the game design stability, uniquely associates and binds the scene identifier with the corresponding progress information, and creates a game scene identifier progress information mapping function;

[0061] The feedback data division unit obtains the player's playing feedback data in each game scene based on the game background database, and obtains the player's playing progress actual operation feedback data array according to the scene identifier;

[0062] The mapping function update unit uses the actual operation feedback data array of the player's game progress to search with the scene identifier as the unique key position, updates the game scene identification progress information mapping function, and establishes the game scene identification associated task type progress information mapping matrix B ;

[0063] The normalization unit normalizes each element in the game scene identifier associated task type progress information mapping matrix to obtain the game scene identifier associated task type progress information mapping normalization matrix ;

[0064] A weighting unit, based on an entropy weight method, assigns a weight to each associated task type according to the overall proportion of the normalized value of the progress information of the task type associated with the game scene identifier, and obtains the weight of the task type associated with the game scene identifier;

[0065] The difficulty evaluation unit maps each element in the normalized matrix of the task type progress information associated with the game scene identifier to the weight of the task type associated with the game scene identifier, and calculates the difficulty index of the task type in the game scene.

[0066] Optionally, the personnel screening module includes:

[0067] The matching calculation unit calculates the matching degree between the preferred behavior style index of each player and the difficulty index of the game scene based on cosine similarity, and constructs the game scene matching matrix C of the players;

[0068] The restriction condition unit, based on the game scene matching matrix of the game players, uses the lower limit of the number of team members in each game scene as a restriction condition;

[0069] An objective function unit, which uses the search for a player matching value that maximizes the current game scenario as an objective function;

[0070] The second model building unit builds a game personnel screening model in the game scene;

[0071] The second model analysis unit, based on the game personnel screening model in the game scene, takes each element in the game personnel's game scene matching matrix as input, and outputs the optimal combination of game personnel in the game scene of the game process progress.

[0072] Compared with the prior art, the present invention has the following beneficial effects:

[0073] The present invention proposes a multi-person interaction solution based on a VR environment. By evaluating the playing preference ability of players and correlating and matching it with the task difficulty of the game progress, the player combination that maximizes the VR game task experience is screened out, a guidance path scene is generated, the interaction frequency in the VR game scene is increased, and the game immersion is enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 It is a flow chart of a multi-person interaction method based on VR environment;

[0075] Figure 2 A flow chart of the method for analyzing the preferred behavioral style indicators of each player;

[0076] Figure 3 A flow chart of the method for evaluating the difficulty index of game scenarios;

[0077] Figure 4 A flowchart of the optimal personnel combination method for generating a game scenario progress chart;

[0078] Figure 5 This is a framework diagram of a multi-person interactive system based on a VR environment. DETAILED DESCRIPTION

[0079] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0080] Reference Figure 1 As shown, a multi-person interaction method based on a VR environment includes:

[0081] Based on the test levels, collect the player's playing data in the VR game and build a player database;

[0082] Based on the player database, analyze the preferred behavior style indicators of each player according to the action events in the player data;

[0083] Based on the game design documents, obtain the game scenes corresponding to the real-time progress in the linear process of the game, analyze the task types in the game scenes, and evaluate the difficulty indicators of the game scenes;

[0084] According to the difficulty index of the game scene, the preferred behavior style index of each player is screened for correlation matching to generate the optimal player combination for the game scene of the game process progress;

[0085] According to the optimal personnel combination for the task in the progress of the game process, and based on the current game space coordinate position of the personnel combination, a guidance path scene is generated to complete the game interaction matching.

[0086] It is understandable that since the differences between VR game content and desktop and mobile terminals are highlighted in the interaction and immersive experience, if mask loading is performed after player matching in the real-time game scene, there will be a gameplay vacuum period, resulting in a decrease in the VR immersive experience. Therefore, the matching combination is searched through the player's preferred ability and the difficulty of the task type, and a guidance path is generated to solve the problem of scene separation during matching loading.

[0087] This solution evaluates the playing preferences of players and correlates them with the task difficulty of the game progress, screens out the personnel combination that maximizes the VR game task experience, generates a guidance path scene, increases the frequency of interaction within the VR game scene, and enhances the game immersion.

[0088] Reference Figure 2As shown, based on the player database, according to the action events in the player data, the preferred behavior style indicators of each player are analyzed, which specifically include:

[0089] Based on the action events in the player data, the action events are initially divided according to multi-dimensional directions to form a multi-dimensional data matrix of the players. A ; The multi-dimensional directions include: task exploration events, skill triggering events, and game triggering events;

[0090] ;

[0091] in, For the Player's Dimensional behavioral data parameters, is the total number of players, is the total number of multi-dimensional directional data;

[0092] Based on the game background database, the player's style and behavior feature data and label data are obtained, and the player's style and behavior feature data is used as a multi-dimensional original sample data set of the game players;

[0093] Based on random forest, the multi-dimensional original sample data set of game players is used as input, and the multi-dimensional original samples contained in the label data are used to train the corresponding game multi-dimensional behavior style recognition decision tree. The minimization of the label data classification error function of the multi-dimensional original sample data division result is used as the training goal, and the combination is a game multi-dimensional behavior style recognition model;

[0094] Substitute each element in the multi-dimensional data matrix of the players into the multi-dimensional game behavior style recognition model, use the maximum return Gini index of the multi-dimensional data of the players for the branch and leaf nodes of the multi-dimensional game behavior style recognition model as the segmentation condition, and generate the player's preferred behavior style index;

[0095] The preferred behavior style index of the players is generated as follows:

[0096] ;

[0097] In the formula, is the preferred behavior style indicator of the players, Identify the model function for the multi-dimensional behavior style of the game. A It is a multi-dimensional data matrix of players. For the Player's Dimensional behavioral data parameters, Divide the elements in the multi-dimensional data matrix of the players intok The proportion of data samples in each behavioral style category.

[0098] This solution integrates the action event data in the player database, constructs a multi-dimensional data matrix, and combines the style behavior feature data and label data in the game background database. It uses the random forest algorithm to train the game multi-dimensional behavior style recognition model, substitutes the multi-dimensional data of the players into the model, and generates the player's preferred behavior style index based on the model's maximum return Gini index. The beneficial effect is that it can accurately identify and analyze the preferred behavior style of the players, maximize the screening of teammates who complement the preferred behavior style of the players, and improve the in-game gaming experience.

[0099] Reference Figure 3 As shown, based on the game design document, the game scene corresponding to the real-time progress in the linear process of the game is obtained, the task types in the game scene are analyzed, and the difficulty indicators of the game scene are evaluated, including:

[0100] Based on the game design document, traverse the line process in the stable game design, uniquely associate and bind the scene identifier with the corresponding progress information, and create a game scene identifier progress information mapping function;

[0101] Based on the game backend database, the player's playing feedback data in each game scene is obtained, and according to the scene identifier, the player's playing progress actual operation feedback data array is obtained; the player's playing feedback data includes: level passing time, level failure number, and level failure frequency;

[0102] Using the actual operation feedback data array of the player's game progress, the scene identifier is used as the unique key to search, update the game scene identifier progress information mapping function, and establish the game scene identifier associated task type progress information mapping matrix B ;

[0103] ;

[0104] in, For the The identifier of the game scene corresponds to the The progress information of the associated task type, is the total number of scene identifiers, The total number of progress information of the associated task type;

[0105] Normalize each element in the game scene identifier associated task type progress information mapping matrix to obtain the game scene identifier associated task type progress information mapping normalized matrix ;

[0106] ;

[0107] in, For the The identifier of the game scene corresponds to the Normalized value of progress information of associated task types;

[0108] Based on the entropy weight method, a weight is assigned to each associated task type according to the overall proportion of the normalized value of the progress information of the task type associated with the game scene identifier, and the weight of the task type associated with the game scene identifier is obtained;

[0109] Based on the game scene identifier-associated task type progress information mapping, each element in the normalized matrix is ​​associated with the task type weight associated with the game scene identifier, and the task type difficulty index in the game scene is calculated;

[0110] The calculation of the difficulty index of the task type in the game scene is specifically:

[0111] ;

[0112] In the formula, The first The difficulty index of each game scene, For the The weight of each associated task type.

[0113] This solution creates a mapping relationship between game scenes and progress information by traversing the route flow in the game design document, and updates the mapping relationship based on the actual game feedback data (including level passing time, number of failures and frequency) in the game background database, forming a mapping matrix between game scenes and task type progress information, and assigns weights to different task types through normalization and entropy weight method, and finally calculates the task type difficulty index in the game scene. The beneficial effect is that it can objectively and accurately evaluate the difficulty of each task type in the game, and provide matching reference parameters for subsequent players.

[0114] Reference Figure 4 As shown, the preferred behavior style indicators of each player are selected for correlation matching according to the difficulty index of the game scene, and the optimal player combination of the game scene to generate the game process progress specifically includes:

[0115] Based on cosine similarity, the matching degree between the preferred behavior style index of each player and the difficulty index of the game scene is calculated to construct the game scene matching matrix C of the players;

[0116] ;

[0117] In the formula, For the The preferred behavior style index of each player is related to the The matching value between the difficulty indicators of the game scenes;

[0118] Based on the game scene matching matrix of the game players, the lower limit of the number of team members in each game scene is used as a restriction condition;

[0119] The objective function is to search for the player matching value that maximizes the current game scenario;

[0120] Construct a model for screening game personnel in the game scene;

[0121] Based on the player screening model in the game scene, each element in the player's game scene matching matrix is ​​used as input, and the optimal player combination of the game scene in the game process progress is used as output;

[0122] The optimal player combination for the game scene of the game process is as follows:

[0123] ;

[0124] In the formula, is a binary decision variable (0,1) when it is 1, Indicates the player Assigned to game scene middle, Indicates the player Not assigned to a game scene middle, It is the lower limit of the number of team members in the game scene. Assign unique constraints to people, Assign unique player count constraints to game scenarios.

[0125] It should be noted that the VR game in this solution is a non-online RPG game. Players all enter the virtual game scene after initially forming a team offline. After being guided and divided according to this solution, the experience effect during the game can be maximized. Therefore, the factor of match rejection is not considered.

[0126] This solution constructs a matching matrix by calculating the cosine similarity between the preferred behavior style index of the players and the difficulty index of the game scene, and combines the team size limit of the game scene to maximize the matching value of the current game scene and build a game personnel screening model. This model can screen out the personnel combination that best matches the difficulty of the game scene, thereby optimizing the progress of the game process. The beneficial effect is to ensure that the most suitable personnel combination can be found for each game scene, improving the game experience and clearance efficiency.

[0127] Reference Figure 5 As shown, based on the same inventive concept of a multi-person interaction method based on a VR environment, a multi-person interaction system based on a VR environment is proposed, including:

[0128] Test database: Based on the test levels, collect the player play data in the VR game and build a player database;

[0129] A player style evaluation module, the style evaluation module is electrically connected to the test database and is based on the player database, and analyzes the preferred behavior style indicators of each player according to the action events in the player data;

[0130] The game difficulty assessment module is used to obtain the game scene corresponding to the real-time progress in the linear process of the game based on the game design document, analyze the task type in the game scene, and evaluate the difficulty index of the game scene;

[0131] A personnel screening module, which is electrically connected to the personnel style assessment module and the game difficulty assessment module. The personnel screening module is used to perform correlation matching screening based on the preferred behavior style indicators of each player according to the difficulty indicators of the game scene, and generate the optimal personnel combination for the game scene of the game process progress;

[0132] The guidance module is electrically connected to the personnel screening module. The guidance module is used to generate a guidance path scene based on the optimal personnel combination for the task according to the progress of the game process and the current game space coordinate position of the personnel combination to complete the game interaction matching.

[0133] The personnel style assessment module includes:

[0134] The multi-dimensional matrix unit performs initial division of the action events according to the multi-dimensional directions based on the action events in the player data, and forms a multi-dimensional data matrix of the players;

[0135] The feedback big data unit obtains the player's style behavior feature data and label data based on the game background database, and uses the player's style behavior feature data as a multi-dimensional original sample data set of the game player;

[0136] The first model building unit, based on random forest, takes the multi-dimensional original sample data set of game players as input, takes the multi-dimensional original samples contained in the label data, trains the corresponding game multi-dimensional behavior style recognition decision tree, takes the minimization of the label data classification error function of the multi-dimensional original sample data division result as the training goal, and combines them into a game multi-dimensional behavior style recognition model;

[0137] The first model analysis unit substitutes each element in the multi-dimensional data matrix of the players into the multi-dimensional behavior style recognition model of the game, and uses the maximum return Gini index of the multi-dimensional data of the players for the branch and leaf nodes of the multi-dimensional behavior style recognition model of the game as a segmentation condition to generate the player's preferred behavior style index.

[0138] Among them, the game difficulty assessment module includes:

[0139] A mapping function unit, based on the game design document, traverses the line flow in the game design stability, uniquely associates and binds the scene identifier with the corresponding progress information, and creates a game scene identifier progress information mapping function;

[0140] The feedback data division unit obtains the player's playing feedback data in each game scene based on the game background database, and obtains the player's playing progress actual operation feedback data array according to the scene identifier;

[0141] The mapping function update unit uses the actual operation feedback data array of the player's game progress to search with the scene identifier as the unique key position, updates the game scene identification progress information mapping function, and establishes the game scene identification associated task type progress information mapping matrix B ;

[0142] A normalization unit, performing normalization processing on each element in the game scene identifier associated task type progress information mapping matrix to obtain a game scene identifier associated task type progress information mapping normalization matrix;

[0143] A weighting unit, based on an entropy weight method, assigns a weight to each associated task type according to the overall proportion of the normalized value of the progress information of the task type associated with the game scene identifier, and obtains the weight of the task type associated with the game scene identifier;

[0144] The difficulty evaluation unit maps each element in the normalized matrix of the task type progress information associated with the game scene identifier to the weight of the task type associated with the game scene identifier, and calculates the difficulty index of the task type in the game scene.

[0145] The personnel screening module includes:

[0146] The matching calculation unit calculates the matching degree between the preferred behavior style index of each player and the difficulty index of the game scene based on cosine similarity, and constructs the game scene matching matrix C of the players;

[0147] The restriction condition unit, based on the game scene matching matrix of the game players, uses the lower limit of the number of team members in each game scene as a restriction condition;

[0148] An objective function unit, which uses the search for a player matching value that maximizes the current game scenario as an objective function;

[0149] The second model building unit builds a game personnel screening model in the game scene;

[0150] The second model analysis unit, based on the game personnel screening model in the game scene, takes each element in the game personnel's game scene matching matrix as input, and outputs the optimal combination of game personnel in the game scene of the game process progress.

[0151] The use process of a multi-person interactive system based on a VR environment is as follows:

[0152] Step 1: Based on the test level, collect the player play data in the VR game and build a player database;

[0153] Step 2: Based on the action events in the player data, the action events are initially divided according to multi-dimensional directions to form a multi-dimensional data matrix of the players;

[0154] Step 3: Based on the game background database, obtain the player's style behavior feature data and label data, and use the player's style behavior feature data as a multi-dimensional original sample data set of the game player;

[0155] Step 4: Based on random forest, take the multi-dimensional original sample data set of game players as input, take the multi-dimensional original samples contained in the label data, train the corresponding game multi-dimensional behavior style recognition decision tree, take the minimization of the label data classification error function of the multi-dimensional original sample data division result as the training goal, and combine them into a game multi-dimensional behavior style recognition model;

[0156] Step 5: Substitute each element in the multi-dimensional data matrix of the players into the multi-dimensional game behavior style recognition model, use the maximum return Gini index of the multi-dimensional data of the players for the branch and leaf nodes of the multi-dimensional game behavior style recognition model as the segmentation condition, and generate the player's preferred behavior style index;

[0157] Step 6: Based on the game design document, traverse the line flow in the stable game design, uniquely associate and bind the scene identifier with the corresponding progress information, and create a game scene identifier progress information mapping function;

[0158] Step 7: Based on the game background database, obtain the player's playing feedback data in each game scene, and obtain the player's playing progress actual operation feedback data array according to the scene identifier;

[0159] Step 8: Use the player's actual operation feedback data array to search with the scene identifier as the unique key, update the game scene identifier progress information mapping function, and establish the game scene identifier associated task type progress information mapping matrix B ;

[0160] Step 9: Normalize each element in the game scene identifier associated task type progress information mapping matrix to obtain the game scene identifier associated task type progress information mapping normalized matrix ;

[0161] Step 10: Based on the entropy weight method, a weight is assigned to each associated task type according to the overall proportion of the normalized value of the progress information of the task type associated with the game scene identifier, and the weight of the task type associated with the game scene identifier is obtained;

[0162] Step 11: Map each element in the normalized matrix based on the progress information of the task type associated with the game scene identifier and the weight of the task type associated with the game scene identifier, and calculate the difficulty index of the task type in the game scene;

[0163] Step 12: Based on the cosine similarity, the matching degree between the preferred behavior style index of each player and the difficulty index of the game scene is calculated, and the game scene matching matrix C of the players is constructed;

[0164] Step 13: Based on the game scene matching matrix of the game players, the lower limit of the number of team members in each game scene is used as a restriction condition;

[0165] Step 14: Taking the search for the maximum player matching value that satisfies the current game scenario as the objective function;

[0166] Step 15: Construct a model for screening game personnel in the game scene;

[0167] Step 16: Based on the game personnel screening model in the game scene, each element in the game personnel's game scene matching matrix is ​​used as input, and the optimal game personnel combination of the game process progress is used as output.

[0168] In summary, the advantages of the present invention are: improving the interaction frequency in the VR game scene and enhancing the game immersion.

[0169] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A multi-person interaction method based on a VR environment, characterized in that: include: Based on the test levels, collect the player's playing data in the VR game and build a player database; Based on the player database, analyze the preferred behavior style indicators of each player according to the action events in the player data; Based on the game design documents, obtain the game scenes corresponding to the real-time progress in the linear process of the game, analyze the task types in the game scenes, and evaluate the difficulty indicators of the game scenes; According to the difficulty index of the game scene, the preferred behavior style index of each player is screened for correlation matching to generate the optimal player combination for the game scene of the game process progress; According to the optimal personnel combination for the task in the game process progress, the guidance path scene is generated according to the current game space coordinate position of the personnel combination to complete the game interaction matching; Among them, based on the player database, according to the action events in the player data, the preferred behavior style indicators of each player are analyzed, including: Based on the action events in the player data, the action events are initially divided according to multi-dimensional directions to form a multi-dimensional data matrix of the players. A ; The multi-dimensional directions include: task exploration events, skill triggering events, and game triggering events; ； in, For the Player's Dimensional behavioral data parameters, is the total number of players, is the total number of multi-dimensional directional data; Based on the game background database, the player's style and behavior feature data and label data are obtained, and the player's style and behavior feature data is used as a multi-dimensional original sample data set of the game players; Based on random forest, the multi-dimensional original sample data set of game players is used as input, and the multi-dimensional original samples contained in the label data are used to train the corresponding game multi-dimensional behavior style recognition decision tree. The minimization of the label data classification error function of the multi-dimensional original sample data division result is used as the training goal, and the combination is a game multi-dimensional behavior style recognition model; Substitute each element in the multi-dimensional data matrix of the players into the multi-dimensional game behavior style recognition model, use the maximum return Gini index of the multi-dimensional data of the players for the branch and leaf nodes of the multi-dimensional game behavior style recognition model as the segmentation condition, and generate the player's preferred behavior style index; The preferred behavior style index of the players is generated as follows: ； In the formula, It is the preferred behavior style index of the players. Identify the model function for the multi-dimensional behavior style of the game. A It is a multi-dimensional data matrix of players. For the Player's Dimensional behavioral data parameters, Divide the elements in the multi-dimensional data matrix of the players into k The proportion of data samples in each behavioral style category.

2. A multi-person interaction method based on a VR environment according to claim 1, characterized in that: Based on the game design document, obtain the game scene corresponding to the real-time progress in the linear process of the game, analyze the task types in the game scene, and evaluate the difficulty indicators of the game scene, including: Based on the game design document, traverse the line process in the stable game design, uniquely associate and bind the scene identifier with the corresponding progress information, and create a game scene identifier progress information mapping function; Based on the game backend database, the player's playing feedback data in each game scene is obtained, and according to the scene identifier, the player's playing progress actual operation feedback data array is obtained; the player's playing feedback data includes: level passing time, level failure number, and level failure frequency; Using the actual operation feedback data array of the player's game progress, the scene identifier is used as the unique key to search, update the game scene identifier progress information mapping function, and establish the game scene identifier associated task type progress information mapping matrix B ; ； in, For the The identifier of the game scene corresponds to the The progress information of the associated task type, is the total number of scene identifiers, The total number of progress information of the associated task type; Normalize each element in the game scene identifier associated task type progress information mapping matrix to obtain the game scene identifier associated task type progress information mapping normalized matrix ; ； in, For the The identifier of the game scene corresponds to the Normalized value of progress information of associated task types; Based on the entropy weight method, a weight is assigned to each associated task type according to the overall proportion of the normalized value of the progress information of the task type associated with the game scene identifier, and the weight of the task type associated with the game scene identifier is obtained; Based on the game scene identifier-associated task type progress information mapping, each element in the normalized matrix is ​​associated with the task type weight associated with the game scene identifier, and the task type difficulty index within the game scene is calculated; The calculation of the difficulty index of the task type in the game scene is specifically: ； In the formula, The first The difficulty index of each game scene, For the The weight of each associated task type.

3. A multi-person interaction method based on a VR environment according to claim 2, characterized in that: According to the difficulty index of the game scene, the preferred behavior style index of each player is screened for correlation matching, and the optimal player combination of the game scene to generate the game process progress includes: Based on cosine similarity, the matching degree between the preferred behavior style index of each player and the difficulty index of the game scene is calculated to construct the game scene matching matrix C of the players; ； In the formula, For the The preferred behavior style index of each player is related to the The matching value between the difficulty indicators of the game scenes; Based on the game scene matching matrix of the game players, the lower limit of the number of team members in each game scene is used as a restriction condition; The objective function is to search for the player matching value that maximizes the current game scenario; Construct a model for screening game personnel in the game scene; Based on the player screening model in the game scene, each element in the player's game scene matching matrix is ​​used as input, and the optimal player combination of the game scene in the game process progress is used as output; The optimal player combination for the game scene of the game process is as follows: ； In the formula, is a binary decision variable (0,1) when it is 1, Indicates the player Assigned to game scene middle, Indicates the player Not assigned to a game scene middle, It is the lower limit of the number of team members in the game scene. Assign unique constraints to people, Assign unique player count constraints to game scenarios.

4. A multi-person interactive system based on a VR environment, characterized in that: A method for implementing a multi-person interaction method based on a VR environment as claimed in any one of claims 1 to 3, comprising: Test database: Based on the test levels, collect the player play data in the VR game and build a player database; A player style evaluation module, the style evaluation module is electrically connected to the test database and is based on the player database, and analyzes the preferred behavior style indicators of each player according to the action events in the player data; The game difficulty assessment module is used to obtain the game scene corresponding to the real-time progress in the linear process of the game based on the game design document, analyze the task type in the game scene, and evaluate the difficulty index of the game scene; A personnel screening module, which is electrically connected to the personnel style assessment module and the game difficulty assessment module. The personnel screening module is used to perform correlation matching screening based on the preferred behavior style indicators of each player according to the difficulty indicators of the game scene, and generate the optimal personnel combination for the game scene of the game process progress; The guidance module is electrically connected to the personnel screening module. The guidance module is used to generate a guidance path scene based on the optimal personnel combination for the task according to the progress of the game process and the current game space coordinate position of the personnel combination to complete the game interaction matching.

5. The multi-person interactive system based on VR environment according to claim 4, characterized in that: The personnel style assessment module includes: The multi-dimensional matrix unit performs initial division of the action events according to the multi-dimensional directions based on the action events in the player data, and forms a multi-dimensional data matrix of the players; The feedback big data unit obtains the player's style behavior feature data and label data based on the game background database, and uses the player's style behavior feature data as a multi-dimensional original sample data set of the game player; The first model building unit, based on random forest, takes the multi-dimensional original sample data set of game players as input, takes the multi-dimensional original samples contained in the label data, trains the corresponding game multi-dimensional behavior style recognition decision tree, takes the minimization of the label data classification error function of the multi-dimensional original sample data division result as the training goal, and combines them into a game multi-dimensional behavior style recognition model; The first model analysis unit substitutes each element in the multi-dimensional data matrix of the players into the multi-dimensional behavior style recognition model of the game, and uses the maximum return Gini index of the multi-dimensional data of the players for the branch and leaf nodes of the multi-dimensional behavior style recognition model of the game as a segmentation condition to generate the player's preferred behavior style index.

6. The multi-person interactive system based on VR environment according to claim 4, characterized in that: The game difficulty assessment module includes: A mapping function unit, based on the game design document, traverses the line flow in the game design stability, uniquely associates and binds the scene identifier with the corresponding progress information, and creates a game scene identifier progress information mapping function; The feedback data division unit obtains the player's playing feedback data in each game scene based on the game background database, and obtains the player's playing progress actual operation feedback data array according to the scene identifier; The mapping function update unit uses the actual operation feedback data array of the player's game progress to search with the scene identifier as the unique key position, updates the game scene identification progress information mapping function, and establishes the game scene identification associated task type progress information mapping matrix B ; The normalization unit normalizes each element in the game scene identifier associated task type progress information mapping matrix to obtain the game scene identifier associated task type progress information mapping normalization matrix ; A weighting unit, based on an entropy weight method, assigns a weight to each associated task type according to the overall proportion of the normalized value of the progress information of the task type associated with the game scene identifier, and obtains the weight of the task type associated with the game scene identifier; The difficulty evaluation unit maps each element in the normalized matrix of the task type progress information associated with the game scene identifier to the weight of the task type associated with the game scene identifier, and calculates the difficulty index of the task type in the game scene.

7. The multi-person interactive system based on a VR environment according to claim 4, characterized in that: The personnel screening module includes: The matching calculation unit calculates the matching degree between the preferred behavior style index of each player and the difficulty index of the game scene based on cosine similarity, and constructs the game scene matching matrix C of the players; The restriction condition unit, based on the game scene matching matrix of the game players, uses the lower limit of the number of team members in each game scene as a restriction condition; An objective function unit, which uses the search for a player matching value that maximizes the current game scenario as an objective function; The second model building unit builds a game personnel screening model in the game scene; The second model analysis unit, based on the game personnel screening model in the game scene, takes each element in the game personnel's game scene matching matrix as input, and outputs the optimal combination of game personnel in the game scene of the game process progress.

Images