Data processing method and device

Abstract

The embodiment of the present application provides a data processing method and device, which obtains current state information through a scene information extraction module and trains a state evaluation model based on historical interaction data. The held element information, displayed element information and remaining quantity information are innovatively converted into combination features, counting features and time series features respectively, and priority screening is performed through a multi-layer strategy execution unit to comprehensively calculate the combination probability, prediction probability and transfer probability. At the same time, a main-standby interactive operation instruction mechanism is established, and when the main instruction verification fails, it can quickly switch to a suitable alternative plan. This method breaks through the limitations of traditional fixed rule decision-making, realizes intelligent interactive operation decision-making, and provides reliable technical support for virtual combat systems.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a data processing method and device. Background Art

[0002] Online games refer to individual multiplayer online games with a high degree of sustainability that use the Internet as a transmission medium, game operator servers and user computers as processing terminals, and game client software as an information exchange window, with the goal of achieving entertainment, leisure, communication, and virtual achievements.

[0003] In virtual combat systems, traditional interactive operation decisions often rely on fixed rules and simple state judgments, which are difficult to adapt to complex and changing combat environments. Existing technologies generally have problems such as a single decision-making mechanism and poor strategy adaptability, and cannot effectively use historical interactive data for intelligent decision-making. Especially in combat scenarios that require rapid response, simple rule matching methods are difficult to ensure the accuracy and timeliness of operations.

[0004] At the same time, existing systems lack effective feature extraction and combination mechanisms when processing multivariate interactive information, and often use single-dimensional feature analysis, resulting in insufficient reliability of decision results. Traditional methods also fail to fully consider the temporal correlation of interactive operations, making it difficult to capture the dynamic changes in the battle process.

[0005] In addition, current technologies also have deficiencies in the verification and fault-tolerance processing of interactive operations. There is a lack of a complete alternative solution management mechanism. Once the verification of the main operation instruction fails, it is often impossible to switch to the appropriate alternative solution in time, affecting the overall operation efficiency of the system. Solving these problems is of great significance to improving the intelligence level and user experience of the virtual battle system. Summary of the invention

[0006] In response to the problems in the prior art, the present application provides a data processing method and device that can break through the limitations of traditional fixed rule decision-making, realize intelligent interactive operation decision-making, and provide reliable technical support for the virtual combat system.

[0007] In order to solve at least one of the above problems, the present application provides the following technical solutions:

[0008] In a first aspect, the present application provides a data processing method, comprising:

[0009] The current state information of the virtual battle system is obtained through the scene information extraction module, and the current state information includes the information of the held interactive elements, the information of the system displayed interactive elements, and the number of the system remaining interactive elements; state information samples are collected during multiple rounds of virtual battles, and historical interactive data are extracted from the state information samples; the historical interactive data is input into the neural network training to obtain a state evaluation model; the state evaluation model is used to process the held interactive element information to obtain the combination feature, the system displayed interactive element information to obtain the counting feature, and the system remaining interactive element number information to obtain the time series feature;

[0010] The combination feature, the counting feature and the time series feature are input into a strategy combination module, and a plurality of strategy execution units sorted by a predetermined priority are sequentially executed in the strategy combination module; in each of the strategy execution units, a combination probability is calculated in combination with the combination feature, a prediction probability is calculated in combination with the counting feature, and a transition probability is calculated in combination with the time series feature; and a screening result of each of the strategy execution units is used as a screening object of the next strategy execution unit until a final screening result is obtained;

[0011] The final screening results are sorted according to the combined probability, the predicted probability and the transfer probability to generate an interactive operation instruction sequence including a main interactive operation instruction and alternative interactive operation instructions; the interactive operation instruction sequence is input into a system execution module, and the system execution module verifies the legitimacy of the main interactive operation instruction based on preset verification rules; when the verification of the main interactive operation instruction fails, the alternative interactive operation instruction is verified according to the sorting result until the verification passes and the verified interactive operation instruction is executed.

[0012] Furthermore, the collecting of state information samples during multiple rounds of virtual battles, extracting historical interaction data from the state information samples; and inputting the historical interaction data into a neural network training to obtain a state evaluation model include:

[0013] Based on the reinforcement learning strategy, the battle process of the virtual battle system is periodically triggered, and the distribution state, continuous click interval, trigger operation event and interaction feedback result of the interactive elements are recorded in real time during the battle process, and the distribution state, the continuous click interval, the trigger operation event and the interaction feedback result are constructed as a state information sample, and the historical interaction data is extracted from the state information sample using a sliding window mechanism;

[0014] A neural network structure including a convolutional layer, a pooling layer and a fully connected layer is constructed. A plurality of convolution kernels of different sizes are set in the convolutional layer to extract local features of the historical interaction data. The local features are subjected to dimensionality reduction processing by the pooling layer to obtain a feature map. The feature map is mapped to a feature vector space by using the fully connected layer to obtain the parameters of the state assessment model.

[0015] Further, the using of the state assessment model to respectively process the held interactive element information to obtain a combination feature, process the system displayed interactive element information to obtain a count feature, and process the system remaining interactive element quantity information to obtain a time series feature includes:

[0016] The held interactive element information is subjected to feature extraction by the convolutional layer of the state assessment model to obtain the positional relationship and combination relationship between the interactive elements, the positional relationship and combination relationship are dimensionally compressed by the pooling layer, and the combination features characterizing the combination pattern of the interactive elements are obtained by mapping through the fully connected layer;

[0017] The system display interactive element information and the system remaining interactive element quantity information are input into the state evaluation model, and the element density is calculated based on the attention mechanism to focus on the interactive element distribution in the display area to obtain the counting feature. The recurrent neural network structure is applied to the remaining interactive element quantity information to calculate the time series change trend of the interactive state to obtain the time series feature.

[0018] Furthermore, the combination feature, the counting feature and the time series feature are input into a strategy combination module, and a plurality of strategy execution units sorted by a predetermined priority are sequentially executed in the strategy combination module, including:

[0019] Constructing a hierarchical structure of a strategy combination module, setting a plurality of strategy execution units in the strategy combination module in descending order of priority, wherein the strategy execution units include an advantage strategy unit, a matching strategy unit, and a supplementary strategy unit, and inputting the combination feature, the counting feature, and the timing feature into corresponding strategy execution units respectively;

[0020] A priority scoring mechanism is set in the policy combination module, and a dynamic weight is assigned to each policy execution unit based on the priority scoring mechanism. The dynamic weight is adaptively adjusted according to the policy execution effect of the interaction process, and the execution order of each policy execution unit is determined according to the dynamic weight.

[0021] Furthermore, in each of the strategy execution units, the combination probability is calculated in combination with the combination feature, the prediction probability is calculated in combination with the counting feature, and the transition probability is calculated in combination with the time series feature; the screening result of each of the strategy execution units is used as the screening object of the next strategy execution unit until the final screening result is obtained, including:

[0022] A three-way parallel probability calculation network is constructed in each strategy execution unit. The first way inputs the combined features into a fully connected layer to generate a feature vector, and the combined probability of different interaction element combinations is calculated by a softmax function. The second way applies a bidirectional long short-term memory network to the counting features to calculate the predicted probability of the interaction effect. The third way uses a variational autoencoder to process the time series features to generate a state transition probability distribution.

[0023] A cascade screening mechanism is constructed, and a screening threshold is set based on the combination probability, the prediction probability and the transfer probability. High-priority interaction element combinations are screened in the advantage strategy unit, and the screening results are passed to the matching strategy unit for secondary screening. Finally, the supplementary strategy unit performs supplementary screening in combination with historical interaction data to obtain the final screening results.

[0024] Furthermore, the final screening results are sorted according to the combination probability, the prediction probability and the transition probability to generate an interactive operation instruction sequence including a main interactive operation instruction and alternative interactive operation instructions, including:

[0025] Constructing a multi-objective optimization function, setting the combination probability as the main optimization objective, setting the prediction probability and the transition probability as constraints, calculating a comprehensive score for each combination of interactive elements in the final screening result, and sorting the interactive element combinations in descending order based on the comprehensive score;

[0026] The interactive element combination with the highest comprehensive score in the sorting result is encoded as the main interactive operation instruction, the interactive element combination with the top N comprehensive scores is selected and encoded as the alternative interactive operation instructions, and the main interactive operation instructions and the alternative interactive operation instructions are organized into an interactive operation instruction sequence in the order of scores using an instruction queue structure.

[0027] Further, the interactive operation instruction sequence is input into a system execution module, the system execution module verifies the legitimacy of the main interactive operation instruction based on a preset verification rule, and when the verification of the main interactive operation instruction fails, verifies the alternative interactive operation instruction according to the sorting result until the verification passes and executes the verified interactive operation instruction, including:

[0028] Constructing a verification rule set for the system execution module, the verification rule set includes interactive element availability verification, state consistency verification and system resource verification, inputting the main interactive operation instruction into the verification rule set, performing multi-level pipeline verification in order of rule priority, and generating a verification result identifier;

[0029] An automatic switching mechanism is set in the verification rule set. When the verification result indicates that the verification of the main interactive operation instruction fails, the alternative interactive operation instructions are sequentially retrieved from the interactive operation instruction sequence to enter the verification process until an alternative interactive operation instruction passes the verification. The verified interactive operation instruction is then submitted to the system execution queue to execute the interactive operation.

[0030] In a second aspect, the present application provides a data processing device, including:

[0031] The interactive element processing module is used to obtain the current state information of the virtual battle system through the scene information extraction module, wherein the current state information includes the held interactive element information, the system displayed interactive element information and the number of interactive elements remaining in the system; collect state information samples during multiple rounds of virtual battles, and extract historical interactive data from the state information samples; input the historical interactive data into the neural network training to obtain a state evaluation model; use the state evaluation model to process the held interactive element information to obtain the combination feature, process the system displayed interactive element information to obtain the counting feature, and process the system remaining interactive element number information to obtain the time series feature;

[0032] A strategy screening module, used for inputting the combination feature, the counting feature and the timing feature into a strategy combination module, and sequentially executing a plurality of strategy execution units sorted by a predetermined priority in the strategy combination module; calculating the combination probability in combination with the combination feature, calculating the prediction probability in combination with the counting feature, and calculating the transition probability in combination with the timing feature in each of the strategy execution units; and using the screening result of each of the strategy execution units as the screening object of the next strategy execution unit until a final screening result is obtained;

[0033] The intelligent recommendation module is used to sort the final screening results according to the combination probability, the predicted probability and the transition probability, and generate an interactive operation instruction sequence including a main interactive operation instruction and an alternative interactive operation instruction; the interactive operation instruction sequence is input into the system execution module, and the system execution module verifies the legitimacy of the main interactive operation instruction based on a preset verification rule. When the verification of the main interactive operation instruction fails, the alternative interactive operation instruction is verified according to the sorting result until the verification passes and the verified interactive operation instruction is executed.

[0034] In a third aspect, the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the data processing method when executing the program.

[0035] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the data processing method when executed by a processor.

[0036] In a fifth aspect, the present application provides a computer program product, comprising a computer program / instructions, which implement the steps of the data processing method when executed by a processor.

[0037] It can be seen from the above technical solution that the present application provides a data processing method and device, which obtains current state information through a scene information extraction module, and trains a state evaluation model based on historical interaction data. The holding element information, display element information and remaining quantity information are innovatively converted into combination features, counting features and timing features respectively, and priority screening is performed through a multi-layer strategy execution unit to comprehensively calculate the combination probability, prediction probability and transfer probability. At the same time, a master-slave interactive operation instruction mechanism is established, and when the main instruction verification fails, it can quickly switch to a suitable alternative plan. This method breaks through the limitations of traditional fixed rule decision-making, realizes intelligent interactive operation decision-making, and provides reliable technical support for virtual combat systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 This is one of the flow charts of the data processing method in the embodiment of the present application;

[0040] Figure 2 This is a second flow chart of the data processing method in an embodiment of the present application;

[0041] Figure 3 This is a third flow chart of the data processing method in the embodiment of the present application;

[0042] Figure 4 This is a fourth flowchart of the data processing method in an embodiment of the present application;

[0043] Figure 5 This is a fifth flow chart of the data processing method in the embodiment of the present application;

[0044] Figure 6 This is a sixth flow chart of the data processing method in the embodiment of the present application;

[0045] Figure 7FIG7 is a flowchart of the data processing method in the embodiment of the present application;

[0046] Figure 8 A structural diagram of a data processing device in an embodiment of the present application;

[0047] Fig. 9 It is a schematic diagram of the structure of an electronic device in an embodiment of the present application.

[0048] Reference numerals:

[0049] Electronic device 9600, central processing unit 9100, memory 9140, communication module 9110, input unit 9120, audio processor 9130, display 9160, power supply 9170, buffer memory 9141, application / function storage unit 9142, data storage unit 9143, driver program storage unit 9144, antenna 9111, speaker 9131, microphone 9132. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0051] The acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of national laws and regulations.

[0052] Taking into account the problems existing in the prior art, the present application provides a data processing method and device, which obtains the current state information through the scene information extraction module, and trains the state evaluation model based on the historical interaction data. The holding element information, display element information and remaining quantity information are innovatively converted into combination features, counting features and timing features respectively, and priority screening is performed through a multi-layer strategy execution unit to comprehensively calculate the combination probability, prediction probability and transfer probability. At the same time, a main-standby interactive operation instruction mechanism is established, and when the main instruction verification fails, it can quickly switch to a suitable alternative plan. This method breaks through the limitations of traditional fixed rule decision-making, realizes intelligent interactive operation decision-making, and provides reliable technical support for the virtual combat system.

[0053] In order to break through the limitations of traditional fixed rule decision-making, realize intelligent interactive operation decision-making, and provide reliable technical support for the virtual battle system, this application provides an embodiment of a data processing method, see Figure 1 , the data processing method specifically includes the following contents:

[0054] Step S101: obtaining the current state information of the virtual battle system through the scene information extraction module, the current state information including the held interactive element information, the system displayed interactive element information and the number of interactive elements remaining in the system; collecting state information samples during multiple rounds of virtual battles, extracting historical interactive data from the state information samples; inputting the historical interactive data into a neural network for training to obtain a state evaluation model; using the state evaluation model to process the held interactive element information to obtain a combination feature, process the system displayed interactive element information to obtain a counting feature, and process the system remaining interactive element number information to obtain a time series feature;

[0055] Optionally, this embodiment provides an intelligent data processing method. First, the current state information in the virtual battle system is obtained in real time through the scene information extraction module, taking the mahjong game as an example, including the player's hand information (holding interactive element information), the card information that has been played on the table (system display interactive element information) and the number of cards remaining in the deck (system remaining interactive element number information).

[0056] While acquiring these basic information, this embodiment constructs a training sample library by collecting a large amount of real game data. Specifically, in each round of virtual battle, the player's card-playing order, operation selections such as eating, touching, and ganging, and the corresponding winning and losing results are recorded, and these data together constitute the state information sample. From these samples, historical interaction data including card type combinations, operation timing, player strategies, etc. are extracted.

[0057] In order to effectively process and utilize these historical data, this embodiment constructs a special neural network structure as a state evaluation model. The model uses a multi-layer convolutional neural network structure and can automatically learn various characteristic patterns in mahjong games. Through training on a large amount of historical game data, the model gradually masters the ability to evaluate the value of different card combinations.

[0058] When processing the current situation information, the state assessment model of this embodiment extracts features from three dimensions: first, the hand information held is analyzed to extract various possible combination features, such as whether there are straights, doubles, and other combinations; second, by analyzing the cards that have been played on the table, counting features such as the remaining number of various types of cards are obtained; finally, combined with the remaining number of cards in the library, the possible subsequent development trend of the game is predicted to form a time series feature.

[0059] Taking a specific scenario as an example, when a player holds a straight combination of 6 barrels, 7 barrels, and 8 barrels, the state evaluation model will consider the number of barrels that have been played and the number of cards remaining in the deck at the same time, and comprehensively evaluate whether to keep this straight combination or make other choices. This multi-dimensional feature extraction method enables this embodiment to more comprehensively understand the current situation and provide a reliable basis for subsequent decision-making.

[0060] This embodiment solves the technical problem that traditional mahjong AI is not comprehensive enough in understanding the situation in this way. By comprehensively considering the characteristics of the three dimensions of hand combination, known card information and the number of remaining cards, the system's ability to understand the mahjong game is significantly improved, making subsequent decisions more in line with actual game needs and better simulating the thinking mode of human players.

[0061] The solution of this embodiment is not only applicable to mahjong games, but can also be extended to other types of chess and card games. By adjusting the dimension of feature extraction and the evaluation criteria, the solution can adapt to the needs of different game types and has strong versatility.

[0062] Step S102: input the combination feature, the counting feature and the time series feature into a strategy combination module, and sequentially execute a plurality of strategy execution units sorted by a predetermined priority in the strategy combination module; calculate the combination probability in combination with the combination feature, calculate the prediction probability in combination with the counting feature, and calculate the transition probability in combination with the time series feature in each strategy execution unit; use the screening result of each strategy execution unit as the screening object of the next strategy execution unit until the final screening result is obtained;

[0063] Optionally, after acquiring various features, this embodiment inputs these features into a strategy combination module for in-depth analysis and decision-making. The module adopts a hierarchical structural design and includes multiple strategy execution units, which are executed in sequence according to a predetermined priority order to form a complete decision chain.

[0064] In the Mahjong game scenario, the strategy execution unit mainly includes the advantage strategy unit, matching strategy unit and supplementary strategy unit. The advantage strategy unit prioritizes high-priority strategies such as listening to cards and winning cards, the matching strategy unit is responsible for processing medium-priority actions such as eating, touching and ganging, and the supplementary strategy unit processes basic strategies such as ordinary playing cards. This hierarchical design ensures the rationality and efficiency of the decision-making process.

[0065] In this embodiment, three types of probability calculations are performed simultaneously in each strategy execution unit. The first is the calculation of the combination probability. For example, in the advantage strategy unit, by analyzing the combination characteristics of the current hand, the probability of various possible ways of listening to the cards is calculated. Taking "1 card 2 cards 3 cards 4 cards 5 cards" as an example, the system will calculate whether to keep the "123" pair to pursue a greater chance of winning, or to disassemble this combination to pursue other possibilities.

[0066] The second is the calculation of prediction probability. By analyzing the counting characteristics, the possible actions of other players are predicted. For example, when a certain type of card has appeared frequently, the system will reduce the prediction probability of the relevant combination to avoid ineffective waiting. The third is the calculation of transfer probability. By analyzing the time series characteristics, the changing trend of the game status is predicted to provide a basis for long-term decision-making.

[0067] This embodiment adopts a cascade screening mechanism, and the screening results of each strategy execution unit will serve as the input of the next unit. For example, when the advantage strategy unit screens out several possible listening plans, these plans will be passed to the matching strategy unit for further screening based on the situation on the field, and finally the supplementary strategy unit will make the final decision based on historical data.

[0068] Through this multi-level strategy combination and screening mechanism, this embodiment effectively solves the technical problem of single decision-making and lack of flexibility of traditional mahjong AI. In practical applications, this solution can adaptively adjust the strategy priority according to different scenarios and make decisions that are more in line with the actual situation.

[0069] This strategy combination method of this embodiment not only improves the accuracy of decision-making, but also significantly improves the adaptability of the system. By dynamically adjusting the weights of each strategy unit, the system can make corresponding strategy adjustments at different game stages, such as preferring to collect card types in the early stage of the game, focusing on defense in the middle stage, and preferring offensive strategies in the later stage.

[0070] The technical effect of this solution is that it is more in line with the decision-making habits of human players in actual games. It can adjust strategies in a timely manner according to changes in the situation, avoid the rigid decision-making problems of traditional AI systems, and significantly improve the intelligence and realism of the game.

[0071] Step S103: Sort the final screening results according to the combination probability, the predicted probability and the transfer probability to generate an interactive operation instruction sequence including a main interactive operation instruction and alternative interactive operation instructions; input the interactive operation instruction sequence into a system execution module, and the system execution module verifies the legitimacy of the main interactive operation instruction based on preset verification rules. When the verification of the main interactive operation instruction fails, verify the alternative interactive operation instruction according to the sorting result until the verification passes and execute the interactive operation instruction that passes the verification.

[0072] Optionally, after completing the strategy screening, this embodiment needs to scientifically and reasonably sort the final screening results to generate an executable interactive operation instruction sequence. Specifically, by comprehensively considering the values ​​of the three dimensions of combination probability, prediction probability and transition probability, a complete decision priority system is established.

[0073] In the Mahjong game scenario, this embodiment first preliminarily sorts possible operations according to the combination probability. For example, when it is detected that both the touch and eat operations can be performed at the same time, the system will give priority to the operation with a higher combination probability. If the probability of the combination of the gong formed after the touch is higher than the probability of the combination of the shun formed after the eat, the touch operation will be ranked higher.

[0074] Next, this embodiment will adjust the preliminary sorting results in combination with the predicted probability. For example, even if the combined probability of a certain card-touching operation is high, if the predicted probability analysis finds that the card type may be needed by other players, the system will reduce the priority of the operation accordingly to avoid creating favorable opportunities for the opponent.

[0075] At the same time, this embodiment will also optimize the operation sequence by analyzing the transition probability. For example, in the late game, even if the current combined probability and predicted probability of an operation are high, if the transition probability shows that the operation may cause the opponent to listen to the card, the system will reduce the priority of the operation and choose a safer strategy instead.

[0076] Through the comprehensive calculation and weighing of the three probabilities, this embodiment finally generates a complete instruction sequence including a main interactive operation instruction and multiple alternative interactive operation instructions. In actual scenarios, if the player's hand contains two sets of straights, "2 tubes 3 tubes 4 tubes" and "60,000 70,000 80,000", the system may use other cards that do not affect the two sets of straights as the main operation instruction, and use the operation of dismantling one of the sets of straights as an alternative instruction.

[0077] After receiving the instruction sequence, the system execution module of this embodiment will first verify the legitimacy of the main interactive operation instruction. The verification rules include multiple dimensions such as whether the operation timing is appropriate and whether the operation complies with the game rules. When the main instruction verification fails, the system will automatically verify the alternative instructions in the established sorting order until a legal and feasible operation instruction is found and executed.

[0078] This design effectively solves the technical problem that the traditional mahjong AI system is prone to freeze or execution errors in the operation execution link. By establishing a complete instruction sequence and verification mechanism, this embodiment significantly improves the system's fault tolerance and operational stability.

[0079] The solution of this embodiment shows good adaptability and reliability in practical applications. By comprehensively analyzing multi-dimensional probabilities for decision sorting and coordinating with a strict verification mechanism, the system can make reasonable operational decisions in various complex scenarios, effectively improving the fluency and realism of the game. At the same time, the versatility of this solution makes it easy to expand to other types of chess and card games, and has a wide range of application value.

[0080] From the above description, it can be seen that the data processing method provided in the embodiment of the present application can obtain the current state information through the scene information extraction module, and train the state evaluation model based on the historical interaction data. The holding element information, display element information and remaining quantity information are innovatively converted into combination features, counting features and timing features respectively, and priority screening is performed through a multi-layer strategy execution unit to comprehensively calculate the combination probability, prediction probability and transfer probability. At the same time, a main-standby interactive operation instruction mechanism is established, and when the main instruction verification fails, it can quickly switch to a suitable alternative plan. This method breaks through the limitations of traditional fixed rule decision-making, realizes intelligent interactive operation decision-making, and provides reliable technical support for the virtual combat system.

[0081] In one embodiment of the data processing method of the present application, see Figure 2 , and can also include the following:

[0082] Step S201: periodically triggering a battle process of a virtual battle system based on a reinforcement learning strategy, recording the distribution state, continuous click interval, trigger operation event, and interaction feedback result of interactive elements in real time during the battle process, constructing the distribution state, the continuous click interval, the trigger operation event, and the interaction feedback result into a state information sample, and extracting historical interaction data from the state information sample using a sliding window mechanism;

[0083] Step S202: construct a neural network structure including a convolutional layer, a pooling layer and a fully connected layer, set a plurality of convolution kernels of different sizes in the convolutional layer to extract local features of the historical interaction data, perform dimensionality reduction processing on the local features through the pooling layer to obtain a feature map, and use the fully connected layer to map the feature map to the feature vector space to obtain the parameters of the state assessment model.

[0084] Optionally, this embodiment first triggers multiple rounds of virtual battles through a reinforcement learning strategy, and uses a periodic sampling method to ensure the continuity and integrity of data collection. In the mahjong game scenario, the distribution state of each player's hand cards, the time interval between players operating mahjong tiles, the triggered operation events such as eating, touching, and playing cards, and the feedback of game results after these operations are recorded in real time.

[0085] This embodiment uses a sliding window mechanism to process the collected data, and the window size can be adjusted according to actual needs. For example, a sliding window containing 10 rounds is set, and each time it slides back one round, so that the state changes and strategy evolution in multiple consecutive rounds can be captured. In this way, the decision-making mode and strategy preference of players at different game stages can be effectively extracted.

[0086] During the data collection phase, this embodiment pays special attention to the player's operation mode. For example, when a player holds a straight such as "1-2-3", the system will record whether the player chooses to keep the straight or split it, as well as the hesitation time when making this choice (reflected by the interval between consecutive clicks). These detailed information is crucial to understanding the player's decision-making thinking.

[0087] In order to effectively process these complex historical interaction data, this embodiment constructs a multi-level neural network structure. In the convolution layer, multiple convolution kernels of different sizes are set to capture feature patterns of different scales. For example, using a 3×3 convolution kernel can extract local card combination features, while a 5×5 convolution kernel can capture a wider range of card distribution features.

[0088] The pooling layer of this embodiment uses the maximum pooling operation to retain the most significant feature information through dimensionality reduction processing. This not only reduces the amount of calculation, but also improves the robustness of the model to noise. For example, when processing the player's operation sequence, the pooling layer can help identify the most critical decision points and filter out some less important intermediate states.

[0089] In the fully connected layer, this embodiment converts the extracted feature graph into a feature vector, and finally obtains the parameters of the state evaluation model through multiple layers of nonlinear transformation. These parameters contain important information such as the value evaluation of different card combinations and the weight distribution of various operation strategies.

[0090] This embodiment solves the technical problem that the traditional Mahjong AI is not deep enough in state understanding and strategy learning in this way. By comprehensively considering the player's operation behavior, time characteristics and decision results, a more complete state evaluation system is established.

[0091] In practical applications, this deep learning-based state assessment model demonstrates strong generalization capabilities. The model can learn the characteristics and strategic patterns of different playing styles from a large amount of game data, and flexibly apply these experiences in new games. For example, when encountering similar cards, the model can quickly evaluate the pros and cons of different operation options based on the experience learned from historical data.

[0092] The solution of this embodiment not only improves the decision-making ability of the AI ​​system, but also makes it show the thinking characteristics closer to human players, greatly enhancing the fun and challenge of the game. At the same time, the design idea of ​​this solution can also be extended to other types of strategy games, and has a wide range of application prospects.

[0093] In one embodiment of the data processing method of the present application, see Figure 3 , and can also include the following:

[0094] Step S301: extracting features of the held interactive element information through the convolutional layer of the state assessment model to obtain positional relationships and combination relationships between the interactive elements, compressing the dimensions of the positional relationships and combination relationships through the pooling layer, and obtaining combination features characterizing the combination pattern of the interactive elements through mapping through the fully connected layer;

[0095] Step S302: Input the system display interactive element information and the system remaining interactive element quantity information into the state assessment model, focus on the interactive element distribution in the display area based on the attention mechanism to calculate the element density to obtain the counting feature, and apply the recurrent neural network structure to the remaining interactive element quantity information to calculate the time series change trend of the interactive state to obtain the time series feature.

[0096] Optionally, this embodiment uses a multi-layer processing mechanism of a state assessment model to perform deep feature extraction on the held interactive element information. In the mahjong game scenario, the mahjong tile information in the player's hand is first input into the convolution layer for processing. The convolution layer analyzes the correlation between each tile and the adjacent tiles by means of a sliding window, and extracts the positional relationship and combination relationship between the tiles.

[0097] In the specific implementation process, the convolution operation of this embodiment can identify a straight relationship such as "1, 2, 3", or a triplet relationship such as "Dong Dong Dong". At the same time, it can also discover potential listening relationships, such as the "13" relationship that may be formed between "1, 2" and "4". Through the superposition of multiple layers of convolution, the system can gradually extract more complex card combination features.

[0098] This embodiment uses a pooling layer to reduce the dimension of the extracted features and retain the most representative feature information. For example, when multiple similar card combinations are identified, the pooling operation can extract the most valuable combination features to avoid information redundancy. After the mapping conversion of the fully connected layer, the combination features that can accurately characterize the current hand combination pattern are finally obtained.

[0099] When processing the interactive element information displayed by the system, this embodiment innovatively introduces an attention mechanism to focus on the distribution of card types in the display area. For example, when a certain card type is found to appear frequently in the display area, the attention mechanism will automatically increase the attention to the card type, thereby more accurately calculating the density distribution of various types of cards.

[0100] This embodiment not only considers the number of cards of each type in the display area, but also their appearance positions and time sequence when calculating element density. For example, when a player plays the same card continuously in a short period of time, the system will pay special attention to this abnormal playing pattern, and this information will be integrated into the counting feature.

[0101] For the information on the number of remaining interactive elements, this embodiment uses a recurrent neural network structure for processing. This structure is particularly suitable for processing data with time-series dependencies and can effectively capture the dynamic change characteristics during the game. For example, by analyzing the changing trend of the number of remaining cards in the card wall, the system can predict the remaining time of the game, thereby adjusting the aggressiveness or conservatism of the strategy.

[0102] This embodiment effectively solves the technical problem that the traditional Mahjong AI is not comprehensive enough in terms of state understanding through this multi-dimensional feature extraction method. Through the synergy of combined features, counting features and time series features, the system can understand the current game state more accurately.

[0103] In practical applications, this feature extraction scheme has shown excellent adaptability and accuracy. For example, when the situation changes significantly, the system can quickly perceive and adjust its strategy. For example, after a player stands upright, the system will immediately re-evaluate the current situation and adjust its defensive strategy.

[0104] This feature extraction scheme of this embodiment not only improves the system's ability to understand the game state, but also enhances the flexibility and accuracy of decision-making. By deeply analyzing the correlation between various features, the system can make judgments that are more in line with the actual situation, significantly improving the intelligence level and playability of the game.

[0105] In one embodiment of the data processing method of the present application, see Figure 4 , and can also include the following:

[0106] Step S401: constructing a hierarchical structure of a strategy combination module, setting a plurality of strategy execution units in the strategy combination module in descending order of priority, wherein the strategy execution units include an advantage strategy unit, a matching strategy unit, and a supplementary strategy unit, and inputting the combination feature, the counting feature, and the timing feature into corresponding strategy execution units respectively;

[0107] Step S402: a priority scoring mechanism is set in the policy combination module, and a dynamic weight is assigned to each policy execution unit based on the priority scoring mechanism. The dynamic weight is adaptively adjusted according to the policy execution effect of the interaction process, and the execution order of each policy execution unit is determined according to the dynamic weight.

[0108] Optionally, this embodiment constructs a strategy combination module with clear hierarchy and reasonable structure, and realizes intelligent decision-making process through the cooperation of multiple strategy execution units. In the mahjong game scenario, the advantage strategy unit is mainly responsible for handling obviously advantageous situations, such as listening to cards and listening all the time; the matching strategy unit is responsible for handling conventional card combinations; and the supplementary strategy unit handles special situations and defensive strategies.

[0109] In this embodiment, the combination features obtained in the previous step are input into the advantage strategy unit to identify favorable combinations in the current hand. For example, when it is detected that there are two complete straights of "1 tube 2 tube 3 tube" and "50,000 60,000 70,000" in the hand, the advantage strategy unit will give priority to retaining these ready-made card combinations and formulate corresponding offensive strategies accordingly.

[0110] The counting features are input into the matching strategy unit to analyze the distribution of cards on the field. For example, by analyzing the cards played by other players, the matching strategy unit can determine the safety of certain cards. When it is found that multiple players have played cards of a certain suit, the system will consider the remaining cards of that suit to be relatively safe and can be used to adjust the playing strategy.

[0111] This embodiment inputs the time series features into the supplementary strategy unit to handle special situations in the game process. For example, in the late game, when there are few card walls left, the supplementary strategy unit will pay more attention to defense to avoid the collapse of the situation due to aggressive offense.

[0112] In the strategy combination module, this embodiment sets a flexible priority scoring mechanism to adjust the importance of different strategy units by allocating dynamic weights. This weight is not fixed, but will be adjusted continuously as the game progresses. For example, when a strategy is executed and achieves good results, the system will appropriately increase the weight of the strategy unit; conversely, if the effect of a strategy is not ideal, its weight will be reduced.

[0113] The dynamic weight adjustment process of this embodiment is highly adaptive. For example, in the early stages of the game, the system may give a higher weight to the matching strategy unit, focusing on the construction of the card type; and when it is detected that the opponent may be listening to the card, the system will increase the weight of the supplementary strategy unit and increase the proportion of defense.

[0114] By adjusting the dynamic weights, this embodiment can determine the most reasonable execution order of each strategy execution unit. For example, when the hand is close to being drawn, the advantage strategy unit will get the highest execution priority; and when facing a dangerous situation, the defensive strategy of the supplementary strategy unit will be executed first.

[0115] This design effectively solves the technical problems of the traditional Mahjong AI system, which is that the strategy selection is not flexible and adaptable. Through hierarchical strategy combination and dynamic weight adjustment, the system can make more reasonable decisions according to different situations.

[0116] In practical applications, the solution of this embodiment demonstrates excellent strategic adaptability. The system can flexibly adjust the execution priority of each strategy unit according to the specific situation of the game and make decisions that are more in line with the actual situation. This solution not only improves the decision-making level of the system, but also enhances the intelligence and fun of the game.

[0117] The innovative design of this embodiment enables the system to maintain stable performance in various complex situations, significantly improving the playability and challenge of the game. At the same time, this strategy combination idea can also be extended to other types of strategy games, and has a wide range of application value.

[0118] In one embodiment of the data processing method of the present application, see Figure 5 , and can also include the following:

[0119] Step S501: construct a three-way parallel probability calculation network in each strategy execution unit, wherein the first way inputs the combined features into a fully connected layer to generate a feature vector, and calculates the combined probability of different combinations of interactive elements through a softmax function, the second way applies a bidirectional long short-term memory network to the counting features to calculate the predicted probability of the interaction effect, and the third way uses a variational autoencoder to process the time series features to generate a state transition probability distribution;

[0120] Step S502: Construct a cascade screening mechanism, set a screening threshold based on the combination probability, the prediction probability and the transfer probability, screen high-priority interaction element combinations in the advantage strategy unit, pass the screening results to the matching strategy unit for secondary screening, and finally the supplementary strategy unit performs supplementary screening in combination with historical interaction data to obtain the final screening results.

[0121] Optionally, this embodiment constructs three parallel probability calculation networks in each strategy execution unit to achieve more accurate decision-making through multi-dimensional probability analysis. In the mahjong game scenario, the first network mainly processes the possibility of hand card combinations, the second network predicts the effects of different playing methods, and the third network analyzes the changing trend of the game situation.

[0122] In the first network, this embodiment inputs the combination features into the fully connected layer to generate a feature vector describing the possibility of various card combinations. Through the conversion of the softmax function, the probability distribution of different playing methods is obtained. For example, when there are "20,000, 30,000, 40,000" and "20,000, 30,000" in the hand, the system will calculate the relative advantage probability of keeping a complete straight or breaking it into a pair.

[0123] In this embodiment, a bidirectional long short-term memory network is used in the second network to process the counting feature. This network structure can consider both historical information and future possibilities. For example, by analyzing the card-playing sequence of other players, the system can predict the safety of certain cards, thereby calculating the risks and benefits that different playing methods may bring.

[0124] In the third network, this embodiment innovatively introduces a variational autoencoder to process time series features. This structure can learn the underlying rules of game state transitions and generate a more accurate state transition probability distribution. For example, when a player enters the state of listening to cards, the system can predict the defensive strategies that other players may adopt.

[0125] The cascade screening mechanism constructed in this embodiment gradually optimizes the decision results through a multi-level screening process. First, in the advantage strategy unit, the system sets a higher screening threshold to retain only the most valuable card combinations. For example, when a listening opportunity is detected, the system will give priority to retaining the relevant card combinations.

[0126] In the secondary screening of the matching strategy unit, this embodiment comprehensively considers multiple probability indicators. For example, although a certain card combination performs well in terms of combination probability, it may be screened out at this stage if the predicted probability shows that it has a high risk.

[0127] The supplementary strategy unit of this embodiment performs supplementary screening in the final stage, and at this time, a more comprehensive evaluation will be performed in combination with historical interaction data. For example, if historical data shows that a certain play fails frequently in similar situations, even if the play performs well in the first two rounds of screening, it may be adjusted or replaced.

[0128] This multi-level screening mechanism effectively solves the technical problem that traditional Mahjong AI does not consider all factors in the decision-making process. Through three-way parallel probability calculation and cascade screening process, the system can more comprehensively evaluate the feasibility and pros and cons of each decision.

[0129] In practical applications, the solution of this embodiment demonstrates excellent decision-making capabilities. The system can quickly analyze complex situations and select the most suitable decision for the current situation from many possible options. For example, when faced with multiple choices of listening to cards, the system can comprehensively consider the probability of winning, defensive requirements, and changes in the field, and choose the best way to play.

[0130] The innovative design of this embodiment significantly improves the decision-making accuracy and adaptability of the system. Through multi-dimensional probability analysis and strict screening mechanisms, the system can make reasonable judgments in various complex situations, effectively improving the intelligence level and fun of the game. At the same time, the design idea of ​​this decision-making framework can also be extended to other types of game games, and has broad application prospects.

[0131] In one embodiment of the data processing method of the present application, see Figure 6 , and can also include the following:

[0132] Step S601: construct a multi-objective optimization function, set the combination probability as the main optimization goal, set the prediction probability and the transition probability as constraints, calculate a comprehensive score for each interactive element combination in the final screening result, and sort the interactive element combinations in descending order based on the comprehensive score;

[0133] Step S602: Encode the interactive element combination with the highest comprehensive score in the sorting result as the main interactive operation instruction, select the top N interactive element combinations with the comprehensive scores and encode them as alternative interactive operation instructions, and use the instruction queue structure to organize the main interactive operation instructions and the alternative interactive operation instructions into an interactive operation instruction sequence in the order of the scores.

[0134] Optionally, this embodiment constructs a scientific and reasonable multi-objective optimization function, which achieves a more accurate decision evaluation by comprehensively weighing different probability indicators. In the mahjong game scenario, the combination probability reflects the basic value of the card combination, while the prediction probability and transfer probability serve as important constraints to ensure the safety and feasibility of the selection.

[0135] In the specific implementation of the optimization function, this embodiment uses the combination probability as the main optimization target, which reflects the core importance of the combination of card types in the mahjong game. For example, when there are two complete straights of "1 tube 2 tube 3 tube" and "20,000 30,000 40,000" in the hand, the combination probability of retaining these ready-made card types will obtain a higher weight.

[0136] This embodiment innovatively uses the prediction probability as one of the constraints to evaluate the risk and benefit of a decision. For example, even if the basic value of a certain card combination is high, if the prediction probability shows that the combination may lead to other players listening to the cards, then the comprehensive score of this combination will be subject to corresponding constraints.

[0137] The state transition probability is another important constraint condition, which is used in this embodiment to evaluate the impact of decisions on the development of the situation. For example, when a decision may lead to an unfavorable development of the game state, even if its current combination value is high, it will be negatively constrained by the transition probability.

[0138] This embodiment uses a weighted summation method to integrate various probability indicators when calculating the comprehensive score. Each combination of interactive elements will receive a score that reflects its comprehensive value. This scoring mechanism takes into account factors in multiple dimensions to ensure that the final decision not only meets current needs but also takes into account long-term development.

[0139] Based on the comprehensive score, this embodiment sorts the interactive element combinations in descending order. The highest-scoring combination will be encoded as the main interactive operation instruction, which usually represents the best choice in the current situation. For example, at the moment of listening to the cards, the easiest way to win the game will be encoded as the main instruction.

[0140] This embodiment also selects the top N combined codes as alternative interactive operation instructions, and these alternative solutions provide the system with flexibility in decision-making. For example, when the main interactive operation instruction cannot be executed due to some unexpected circumstances, the system can immediately switch to the alternative solution.

[0141] In constructing the instruction queue, this embodiment uses an orderly structure to organize these instructions. The main interactive operation instruction is at the top of the queue, followed by alternative instructions sorted by score. This structure enables the system to quickly switch decision plans when needed.

[0142] This design effectively solves the technical problem that traditional Mahjong AI is not flexible enough in decision evaluation and solution selection. Through multi-objective optimization and instruction queue design, the system can evaluate each decision more comprehensively and maintain sufficient decision flexibility.

[0143] In practical applications, the solution of this embodiment demonstrates excellent decision-making ability and adaptability. The system can quickly adjust strategies according to changes in the situation, while ensuring the quality of decision-making and maintaining sufficient strategy redundancy. For example, when encountering an unexpected opponent's play, the system can quickly select a suitable response strategy from the alternative solutions.

[0144] The innovative design of this embodiment significantly improves the decision-making accuracy and reliability of the system. Through a scientific scoring mechanism and a flexible command queue, the system can always maintain reasonable judgment in complex and changing situations, effectively improving the intelligence level and fun of the game. At the same time, the design idea of ​​this decision-making framework can also be extended to other types of intelligent game systems, and has a wide range of application value.

[0145] In one embodiment of the data processing method of the present application, see Figure 7 , and can also include the following:

[0146] Step S701: construct a verification rule set for the system execution module, the verification rule set includes interactive element availability verification, state consistency verification and system resource verification, input the main interactive operation instruction into the verification rule set, perform multi-level pipeline verification according to the rule priority order, and generate a verification result identifier;

[0147] Step S702: An automatic switching mechanism is set in the verification rule set. When the verification result indicates that the verification of the main interactive operation instruction fails, alternative interactive operation instructions are sequentially retrieved from the interactive operation instruction sequence to enter the verification process until an alternative interactive operation instruction passes the verification. The verified interactive operation instruction is then submitted to the system execution queue to perform the interactive operation.

[0148] Optionally, this embodiment constructs a complete set of verification rules to ensure the reliability of interactive operations through a multi-level verification mechanism. In the Mahjong game scenario, the verification rule set includes three core verification links: interactive element availability verification, state consistency verification, and system resource verification. These verification links form a multi-level pipeline structure in order of priority.

[0149] In the interactive element availability verification phase, this embodiment first checks whether the selected card type combination is still available. For example, when the system decides to play a card, it needs to verify whether the card is still in the hand and whether it is not locked by other operations. This verification can avoid operation failure caused by card type changes.

[0150] The state consistency verification is the second verification link of this embodiment, which mainly ensures that the current game state is consistent with the state at the time of decision-making. For example, when the system is ready to perform a certain card-push operation, it needs to verify whether it is still in the right time to push the card and whether other players have not performed interruptive operations.

[0151] In the system resource verification phase, this embodiment checks whether the system resources required to perform the operation are sufficient. For example, when processing a complex one-key operation, it is necessary to ensure that the system has sufficient processing power and storage space to complete the entire operation sequence.

[0152] The verification process adopts a pipeline structure, and this embodiment performs verification in sequence according to the priority of the rules. Each verification link will generate a corresponding verification result identifier, and only when all verification links are passed can the operation instruction be submitted for execution. This design ensures the safety and reliability of the operation.

[0153] This embodiment innovatively designs an automatic switching mechanism. When the main interactive operation instruction fails in any verification link, the system will automatically call the next alternative instruction from the instruction sequence to enter the verification process. For example, if the system originally planned to perform a card-touching operation, but the verification found that the current card-touching opportunity has been missed, it will automatically switch to the alternative card-playing operation.

[0154] During the automatic switching process, this embodiment tries the candidate instructions in the previously established scoring order. Each candidate instruction needs to go through the same verification process until a fully verified instruction is found. This mechanism ensures that the system can always find a feasible operation plan.

[0155] When a verified instruction is found, this embodiment submits it to the system execution queue. The execution queue is responsible for executing these operations in a predetermined order and timing to ensure the smoothness and accuracy of the operation.

[0156] This design effectively solves the technical problems that traditional Mahjong AI often encounters when executing decisions, such as operation failure and inconsistent state. Through strict verification mechanisms and flexible switching strategies, the system can better cope with various unexpected situations during the game.

[0157] In practical applications, the verification mechanism of this embodiment significantly improves the reliability and stability of the system. Even in complex situations such as network delays and state changes, the system can maintain the correct operation sequence and avoid erroneous operations or operation failures.

[0158] The innovative design of this embodiment makes the system more robust and flexible when executing decisions. Through multi-level verification and automatic switching mechanisms, the system can maintain high operating efficiency while ensuring operational accuracy. This not only improves the gaming experience, but also provides valuable implementation ideas for other similar intelligent decision-making systems.

[0159] In order to break through the limitations of traditional fixed rule decision-making, realize intelligent interactive operation decision-making, and provide reliable technical support for the virtual battle system, the present application provides an embodiment of a data processing device for realizing all or part of the content of the data processing method, see Figure 8 , the data processing device specifically includes the following contents:

[0160] The interactive element processing module 10 is used to obtain the current state information of the virtual battle system through the scene information extraction module, wherein the current state information includes the held interactive element information, the system displayed interactive element information and the number of interactive elements remaining in the system; collect state information samples during multiple rounds of virtual battles, and extract historical interactive data from the state information samples; input the historical interactive data into a neural network training to obtain a state evaluation model; use the state evaluation model to process the held interactive element information to obtain a combination feature, process the system displayed interactive element information to obtain a counting feature, and process the system remaining interactive element number information to obtain a time series feature;

[0161] The strategy screening module 20 is used to input the combination feature, the counting feature and the timing feature into the strategy combination module, and sequentially execute a plurality of strategy execution units sorted by a predetermined priority in the strategy combination module; in each of the strategy execution units, the combination probability is calculated in combination with the combination feature, the prediction probability is calculated in combination with the counting feature, and the transition probability is calculated in combination with the timing feature; and the screening result of each of the strategy execution units is used as the screening object of the next strategy execution unit until the final screening result is obtained;

[0162] The intelligent recommendation module 30 is used to sort the final screening results according to the combination probability, the predicted probability and the transfer probability, and generate an interactive operation instruction sequence including a main interactive operation instruction and an alternative interactive operation instruction; the interactive operation instruction sequence is input into the system execution module, and the system execution module verifies the legitimacy of the main interactive operation instruction based on a preset verification rule. When the verification of the main interactive operation instruction fails, the alternative interactive operation instruction is verified according to the sorting result until the verification passes and the verified interactive operation instruction is executed.

[0163] From the above description, it can be seen that the data processing device provided in the embodiment of the present application can obtain the current state information through the scene information extraction module, and train the state evaluation model based on the historical interaction data. The holding element information, display element information and remaining quantity information are innovatively converted into combination features, counting features and timing features respectively, and priority screening is performed through a multi-layer strategy execution unit to comprehensively calculate the combination probability, prediction probability and transfer probability. At the same time, a main-standby interactive operation instruction mechanism is established, and when the main instruction verification fails, it can quickly switch to a suitable alternative plan. This method breaks through the limitations of traditional fixed rule decision-making, realizes intelligent interactive operation decision-making, and provides reliable technical support for the virtual combat system.

[0164] From the hardware level, in order to break through the limitations of traditional fixed rule decision-making, realize intelligent interactive operation decision-making, and provide reliable technical support for the virtual battle system, the present application provides an embodiment of an electronic device for implementing all or part of the content of the data processing method, and the electronic device specifically includes the following content:

[0165] Processor, memory, communication interface and bus; wherein the processor, memory and communication interface communicate with each other through the bus; the communication interface is used to realize information transmission between the data processing device and related devices such as the core business system, user terminal and related database; the logic controller can be a desktop computer, a tablet computer and a mobile terminal, etc., but the present embodiment is not limited thereto. In the present embodiment, the logic controller can be implemented with reference to the embodiment of the data processing method and the embodiment of the data processing device in the embodiment, and the contents thereof are incorporated herein, and the repeated parts are not repeated.

[0166] It is understandable that the user terminal may include a smart phone, a tablet electronic device, a network set-top box, a portable computer, a desktop computer, a personal digital assistant (PDA), a vehicle-mounted device, a smart wearable device, etc. Among them, the smart wearable device may include smart glasses, a smart watch, a smart bracelet, etc.

[0167] In practical applications, part of the data processing method can be executed on the electronic device side as described above, or all operations can be completed in the client device. The specific selection can be based on the processing capability of the client device and the limitations of the user's usage scenario. This application does not limit this. If all operations are completed in the client device, the client device may also include a processor.

[0168] The client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side, and other implementation scenarios may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, or a server cluster consisting of multiple servers, or a server structure of a distributed device.

[0169] Fig. 9 FIG. 9 is a schematic block diagram of the system structure of the electronic device 9600 according to an embodiment of the present application. Fig. 9As shown, the electronic device 9600 may include a central processor 9100 and a memory 9140; the memory 9140 is coupled to the central processor 9100. It is worth noting that Fig. 9 is exemplary; other types of structures may also be used to supplement or replace this structure to implement telecommunication functions or other functions.

[0170] In one embodiment, the data processing method function may be integrated into the central processing unit 9100. The central processing unit 9100 may be configured to perform the following control:

[0171] Step S101: obtaining the current state information of the virtual battle system through the scene information extraction module, the current state information including the held interactive element information, the system displayed interactive element information and the number of interactive elements remaining in the system; collecting state information samples during multiple rounds of virtual battles, extracting historical interactive data from the state information samples; inputting the historical interactive data into a neural network for training to obtain a state evaluation model; using the state evaluation model to process the held interactive element information to obtain a combination feature, process the system displayed interactive element information to obtain a counting feature, and process the system remaining interactive element number information to obtain a time series feature;

[0172] Step S102: input the combination feature, the counting feature and the time series feature into a strategy combination module, and sequentially execute a plurality of strategy execution units sorted by a predetermined priority in the strategy combination module; calculate the combination probability in combination with the combination feature, calculate the prediction probability in combination with the counting feature, and calculate the transition probability in combination with the time series feature in each strategy execution unit; use the screening result of each strategy execution unit as the screening object of the next strategy execution unit until the final screening result is obtained;

[0173] Step S103: Sort the final screening results according to the combination probability, the predicted probability and the transfer probability to generate an interactive operation instruction sequence including a main interactive operation instruction and alternative interactive operation instructions; input the interactive operation instruction sequence into a system execution module, and the system execution module verifies the legitimacy of the main interactive operation instruction based on preset verification rules. When the verification of the main interactive operation instruction fails, verify the alternative interactive operation instruction according to the sorting result until the verification passes and execute the interactive operation instruction that passes the verification.

[0174] From the above description, it can be seen that the electronic device provided in the embodiment of the present application obtains the current state information through the scene information extraction module, and trains the state evaluation model based on the historical interaction data. The holding element information, display element information and remaining quantity information are innovatively converted into combination features, counting features and timing features respectively, and priority screening is performed through a multi-layer strategy execution unit to comprehensively calculate the combination probability, prediction probability and transfer probability. At the same time, a main-standby interactive operation instruction mechanism is established, and when the main instruction verification fails, it can quickly switch to a suitable alternative plan. This method breaks through the limitations of traditional fixed rule decision-making, realizes intelligent interactive operation decision-making, and provides reliable technical support for the virtual combat system.

[0175] In another embodiment, the data processing device may be configured separately from the central processing unit 9100. For example, the data processing device may be configured as a chip connected to the central processing unit 9100, and the function of the data processing method may be implemented under the control of the central processing unit.

[0176] like Fig. 9 As shown, the electronic device 9600 may also include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily have to include Fig. 9 In addition, the electronic device 9600 may also include Fig. 9 For components not shown, reference may be made to the prior art.

[0177] like Fig. 9 As shown, the central processing unit 9100 is sometimes also referred to as a controller or an operation control, and may include a microprocessor or other processor device and / or logic device. The central processing unit 9100 receives input and controls the operation of various components of the electronic device 9600.

[0178] The memory 9140 may be, for example, one or more of a cache, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory or other suitable devices. The above-mentioned information related to the failure may be stored, and a program for executing the relevant information may also be stored. The CPU 9100 may execute the program stored in the memory 9140 to implement information storage or processing, etc.

[0179] The input unit 9120 provides input to the central processing unit 9100. The input unit 9120 is, for example, a key or a touch input device. The power supply 9170 is used to provide power to the electronic device 9600. The display 9160 is used to display display objects such as images and texts. The display may be, for example, an LCD display, but is not limited thereto.

[0180] The memory 9140 may be a solid-state memory, such as a read-only memory (ROM), a random access memory (RAM), a SIM card, etc. It may also be a memory that saves information even when the power is off, can be selectively erased, and is provided with more data, examples of which are sometimes referred to as EPROMs, etc. The memory 9140 may also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142, which is used to store application programs and function programs or processes for executing the operation of the electronic device 9600 through the central processor 9100.

[0181] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for communication functions of the electronic device and / or for executing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0182] The communication module 9110 is a transmitter / receiver that sends and receives signals via the antenna 9111. The communication module 9110 (transmitter / receiver) is coupled to the central processor 9100 to provide input signals and receive output signals, which may be the same as the case of a conventional mobile communication terminal.

[0183] Based on different communication technologies, multiple communication modules 9110 may be provided in the same electronic device, such as a cellular network module, a Bluetooth module and / or a wireless LAN module, etc. The communication module 9110 (transmitter / receiver) is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide an audio output via the speaker 9131 and receive an audio input from the microphone 9132, thereby realizing a common telecommunication function. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. In addition, the audio processor 9130 is also coupled to the central processor 9100, so that recording can be performed on the local machine through the microphone 9132, and the sound stored on the local machine can be played through the speaker 9131.

[0184] The embodiments of the present application also provide a computer-readable storage medium capable of implementing all the steps of the data processing method in the above embodiments, where the execution subject is a server or a client. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, all the steps of the data processing method in the above embodiments are implemented by the execution subject being a server or a client. For example, when the processor executes the computer program, the following steps are implemented:

[0185] Step S101: obtaining the current state information of the virtual battle system through the scene information extraction module, the current state information including the held interactive element information, the system displayed interactive element information and the number of interactive elements remaining in the system; collecting state information samples during multiple rounds of virtual battles, extracting historical interactive data from the state information samples; inputting the historical interactive data into a neural network for training to obtain a state evaluation model; using the state evaluation model to process the held interactive element information to obtain a combination feature, process the system displayed interactive element information to obtain a counting feature, and process the system remaining interactive element number information to obtain a time series feature;

[0186] Step S102: input the combination feature, the counting feature and the time series feature into a strategy combination module, and sequentially execute a plurality of strategy execution units sorted by a predetermined priority in the strategy combination module; calculate the combination probability in combination with the combination feature, calculate the prediction probability in combination with the counting feature, and calculate the transition probability in combination with the time series feature in each strategy execution unit; use the screening result of each strategy execution unit as the screening object of the next strategy execution unit until the final screening result is obtained;

[0187] Step S103: Sort the final screening results according to the combination probability, the predicted probability and the transfer probability to generate an interactive operation instruction sequence including a main interactive operation instruction and alternative interactive operation instructions; input the interactive operation instruction sequence into a system execution module, and the system execution module verifies the legitimacy of the main interactive operation instruction based on preset verification rules. When the verification of the main interactive operation instruction fails, verify the alternative interactive operation instruction according to the sorting result until the verification passes and execute the interactive operation instruction that passes the verification.

[0188] From the above description, it can be seen that the computer-readable storage medium provided in the embodiment of the present application obtains the current state information through the scene information extraction module, and trains the state evaluation model based on the historical interaction data. The holding element information, display element information and remaining quantity information are innovatively converted into combination features, counting features and timing features respectively, and priority screening is performed through a multi-layer strategy execution unit to comprehensively calculate the combination probability, prediction probability and transfer probability. At the same time, a main-standby interactive operation instruction mechanism is established, and when the main instruction verification fails, it can quickly switch to a suitable alternative plan. This method breaks through the limitations of traditional fixed rule decision-making, realizes intelligent interactive operation decision-making, and provides reliable technical support for the virtual combat system.

[0189] The embodiments of the present application also provide a computer program product capable of implementing all the steps of the data processing method in the above embodiments, where the execution subject is a server or a client. When the computer program / instruction is executed by a processor, the steps of the data processing method are implemented. For example, the computer program / instruction implements the following steps:

[0190] Step S101: obtaining the current state information of the virtual battle system through the scene information extraction module, the current state information including the held interactive element information, the system displayed interactive element information and the number of interactive elements remaining in the system; collecting state information samples during multiple rounds of virtual battles, extracting historical interactive data from the state information samples; inputting the historical interactive data into a neural network for training to obtain a state evaluation model; using the state evaluation model to process the held interactive element information to obtain a combination feature, process the system displayed interactive element information to obtain a counting feature, and process the system remaining interactive element number information to obtain a time series feature;

[0191] Step S102: input the combination feature, the counting feature and the time series feature into a strategy combination module, and sequentially execute a plurality of strategy execution units sorted by a predetermined priority in the strategy combination module; calculate the combination probability in combination with the combination feature, calculate the prediction probability in combination with the counting feature, and calculate the transition probability in combination with the time series feature in each strategy execution unit; use the screening result of each strategy execution unit as the screening object of the next strategy execution unit until the final screening result is obtained;

[0192] Step S103: Sort the final screening results according to the combination probability, the predicted probability and the transfer probability to generate an interactive operation instruction sequence including a main interactive operation instruction and alternative interactive operation instructions; input the interactive operation instruction sequence into a system execution module, and the system execution module verifies the legitimacy of the main interactive operation instruction based on preset verification rules. When the verification of the main interactive operation instruction fails, verify the alternative interactive operation instruction according to the sorting result until the verification passes and execute the interactive operation instruction that passes the verification.

[0193] From the above description, it can be seen that the computer program product provided by the embodiment of the present application obtains the current state information through the scene information extraction module, and trains the state evaluation model based on the historical interaction data. The holding element information, display element information and remaining quantity information are innovatively converted into combination features, counting features and timing features respectively, and priority screening is performed through a multi-layer strategy execution unit to comprehensively calculate the combination probability, prediction probability and transfer probability. At the same time, a main-standby interactive operation instruction mechanism is established, and when the main instruction verification fails, it can quickly switch to a suitable alternative plan. This method breaks through the limitations of traditional fixed rule decision-making, realizes intelligent interactive operation decision-making, and provides reliable technical support for the virtual combat system.

[0194] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0195] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (apparatus), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0196] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0197] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0198] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A data processing method, characterized in that: The method comprises: The current state information of the virtual battle system is obtained through the scene information extraction module, and the current state information includes the information of the held interactive elements, the information of the system displayed interactive elements, and the number of the system remaining interactive elements; state information samples are collected during multiple rounds of virtual battles, and historical interactive data are extracted from the state information samples; the historical interactive data is input into the neural network training to obtain a state evaluation model; the state evaluation model is used to process the held interactive element information to obtain the combination feature, the system displayed interactive element information to obtain the counting feature, and the system remaining interactive element number information to obtain the time series feature; The combination feature, the counting feature and the time series feature are input into a strategy combination module, and a plurality of strategy execution units sorted by a predetermined priority are sequentially executed in the strategy combination module; in each of the strategy execution units, a combination probability is calculated in combination with the combination feature, a prediction probability is calculated in combination with the counting feature, and a transition probability is calculated in combination with the time series feature; and a screening result of each of the strategy execution units is used as a screening object of the next strategy execution unit until a final screening result is obtained; The final screening results are sorted according to the combined probability, the predicted probability and the transfer probability to generate an interactive operation instruction sequence including a main interactive operation instruction and alternative interactive operation instructions; the interactive operation instruction sequence is input into a system execution module, and the system execution module verifies the legitimacy of the main interactive operation instruction based on preset verification rules; when the verification of the main interactive operation instruction fails, the alternative interactive operation instruction is verified according to the sorting result until the verification passes and the verified interactive operation instruction is executed.

2. The data processing method according to claim 1, characterized in that: The collecting of state information samples during multiple rounds of virtual battles, and extracting historical interaction data from the state information samples; Inputting the historical interaction data into a neural network training to obtain a state assessment model includes: Based on the reinforcement learning strategy, the battle process of the virtual battle system is periodically triggered, and the distribution state, continuous click interval, trigger operation event and interaction feedback result of the interactive elements are recorded in real time during the battle process, and the distribution state, the continuous click interval, the trigger operation event and the interaction feedback result are constructed as a state information sample, and the historical interaction data is extracted from the state information sample using a sliding window mechanism; A neural network structure including a convolutional layer, a pooling layer and a fully connected layer is constructed. A plurality of convolution kernels of different sizes are set in the convolutional layer to extract local features of the historical interaction data. The local features are subjected to dimensionality reduction processing by the pooling layer to obtain a feature map. The feature map is mapped to a feature vector space by using the fully connected layer to obtain the parameters of the state assessment model.

3. The data processing method according to claim 1, characterized in that: The using of the state assessment model to respectively process the held interactive element information to obtain a combination feature, process the system displayed interactive element information to obtain a count feature, and process the system remaining interactive element quantity information to obtain a time series feature includes: The held interactive element information is subjected to feature extraction by the convolutional layer of the state assessment model to obtain the positional relationship and combination relationship between the interactive elements, the positional relationship and combination relationship are dimensionally compressed by the pooling layer, and the combination features characterizing the combination pattern of the interactive elements are obtained by mapping through the fully connected layer; The system display interactive element information and the system remaining interactive element quantity information are input into the state evaluation model, and the element density is calculated based on the attention mechanism to focus on the interactive element distribution in the display area to obtain the counting feature. The recurrent neural network structure is applied to the remaining interactive element quantity information to calculate the time series change trend of the interactive state to obtain the time series feature.

4. The data processing method according to claim 1, characterized in that: The step of inputting the combination feature, the counting feature and the time sequence feature into a strategy combination module, and sequentially executing a plurality of strategy execution units sorted by a predetermined priority in the strategy combination module includes: Constructing a hierarchical structure of a strategy combination module, setting a plurality of strategy execution units in the strategy combination module in descending order of priority, wherein the strategy execution units include an advantage strategy unit, a matching strategy unit, and a supplementary strategy unit, and inputting the combination feature, the counting feature, and the timing feature into corresponding strategy execution units respectively; A priority scoring mechanism is set in the policy combination module, and a dynamic weight is assigned to each policy execution unit based on the priority scoring mechanism. The dynamic weight is adaptively adjusted according to the policy execution effect of the interaction process, and the execution order of each policy execution unit is determined according to the dynamic weight.

5. The data processing method according to claim 1, characterized in that: In each of the strategy execution units, the combination probability is calculated by combining the combination feature, the prediction probability is calculated by combining the counting feature, and the transfer probability is calculated by combining the time series feature; The screening result of each of the policy execution units is used as the screening object of the next policy execution unit until the final screening result is obtained, including: A three-way parallel probability calculation network is constructed in each strategy execution unit. The first way inputs the combined features into a fully connected layer to generate a feature vector, and the combined probability of different combinations of interactive elements is calculated by a softmax function. The second way applies a bidirectional long short-term memory network to the counting features to calculate the predicted probability of the interaction effect. The third way uses a variational autoencoder to process the time series features to generate a state transition probability distribution. A cascade screening mechanism is constructed, and a screening threshold is set based on the combination probability, the prediction probability and the transfer probability. High-priority interaction element combinations are screened in the advantage strategy unit, and the screening results are passed to the matching strategy unit for secondary screening. Finally, the supplementary strategy unit performs supplementary screening in combination with historical interaction data to obtain the final screening results.

6. The data processing method according to claim 1, characterized in that: The final screening results are sorted according to the combination probability, the prediction probability and the transition probability to generate an interactive operation instruction sequence including a main interactive operation instruction and alternative interactive operation instructions, including: Constructing a multi-objective optimization function, setting the combination probability as the main optimization objective, setting the prediction probability and the transition probability as constraints, calculating a comprehensive score for each combination of interactive elements in the final screening result, and sorting the interactive element combinations in descending order based on the comprehensive score; The interactive element combination with the highest comprehensive score in the sorting result is encoded as the main interactive operation instruction, the interactive element combination with the top N comprehensive scores is selected and encoded as the alternative interactive operation instructions, and the main interactive operation instructions and the alternative interactive operation instructions are organized into an interactive operation instruction sequence in the order of scores using an instruction queue structure.

7. The data processing method according to claim 1, characterized in that: The step of inputting the interactive operation instruction sequence into a system execution module, wherein the system execution module verifies the legitimacy of the main interactive operation instruction based on a preset verification rule, and when the verification of the main interactive operation instruction fails, verifying the alternative interactive operation instruction according to the sorting result until the verification passes and executing the verified interactive operation instruction includes: Constructing a verification rule set for the system execution module, the verification rule set includes interactive element availability verification, state consistency verification and system resource verification, inputting the main interactive operation instruction into the verification rule set, performing multi-level pipeline verification in order of rule priority, and generating a verification result identifier; An automatic switching mechanism is set in the verification rule set. When the verification result indicates that the verification of the main interactive operation instruction fails, the alternative interactive operation instructions are sequentially retrieved from the interactive operation instruction sequence to enter the verification process until an alternative interactive operation instruction passes the verification. The verified interactive operation instruction is then submitted to the system execution queue to execute the interactive operation.

8. A data processing device, characterized in that: The device comprises: The interactive element processing module is used to obtain the current state information of the virtual battle system through the scene information extraction module, wherein the current state information includes the held interactive element information, the system displayed interactive element information and the number of interactive elements remaining in the system; collect state information samples during multiple rounds of virtual battles, and extract historical interactive data from the state information samples; input the historical interactive data into the neural network training to obtain a state evaluation model; use the state evaluation model to process the held interactive element information to obtain the combination feature, process the system displayed interactive element information to obtain the counting feature, and process the system remaining interactive element number information to obtain the time series feature; A strategy screening module, used for inputting the combination feature, the counting feature and the timing feature into a strategy combination module, and sequentially executing a plurality of strategy execution units sorted by a predetermined priority in the strategy combination module; calculating the combination probability in combination with the combination feature, calculating the prediction probability in combination with the counting feature, and calculating the transition probability in combination with the timing feature in each of the strategy execution units; and using the screening result of each of the strategy execution units as the screening object of the next strategy execution unit until a final screening result is obtained; The intelligent recommendation module is used to sort the final screening results according to the combination probability, the predicted probability and the transition probability, and generate an interactive operation instruction sequence including a main interactive operation instruction and an alternative interactive operation instruction; the interactive operation instruction sequence is input into the system execution module, and the system execution module verifies the legitimacy of the main interactive operation instruction based on a preset verification rule. When the verification of the main interactive operation instruction fails, the alternative interactive operation instruction is verified according to the sorting result until the verification passes and the verified interactive operation instruction is executed.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the data processing method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the data processing method according to any one of claims 1 to 7 are implemented.

Images