Chess game processing method and device, electronic equipment, computer readable storage medium and computer program product

Abstract

The application provides a processing method and device of a chess game, an electronic device, a computer readable storage medium and a computer program product; the method comprises: obtaining an operation performed by a target object on a chessboard and a chessboard state of the chessboard; performing feature extraction on the operation and the chessboard state respectively to obtain operation features corresponding to the operation and chessboard features corresponding to the chessboard state; determining an object type of the target object based on the operation features and the chessboard features, and determining an agent corresponding to the object type; simulating subsequent operations of the target object on the chessboard through the agent to obtain a pass probability of a current chess piece generation strategy of the chessboard; adjusting the chess piece generation strategy based on the pass probability to obtain an adjusted chess piece generation strategy, and generating a chess piece of the chessboard based on the adjusted chess piece generation strategy, which can improve the generation of a chess piece adapted to the target object.

Description

Technical Field

[0001] This application relates to computer technology, and more particularly to a method, apparatus, electronic device, computer-readable storage medium, and computer program product for processing board games. Background Technology

[0002] In current board games, the strategy for generating pieces on the board is often random. For example, after a user eliminates an existing piece, a functional piece or an obstacle piece is randomly generated. While this allows the user to perform actions normally, the impact of functional pieces and obstacle pieces on the board game is completely different. Therefore, the generated pieces may not match the current state of the board, resulting in the difficulty of completing the game being too high or too low, thus reducing the user's experience. Summary of the Invention

[0003] This application provides a method, apparatus, electronic device, computer-readable storage medium, and computer program product for processing board games, capable of generating chess pieces that conform to the current board state.

[0004] The technical solution of this application embodiment is implemented as follows: This application provides a method for processing a board game, the method comprising: Obtain the operations performed by the target object on the chessboard and the state of the chessboard; Feature extraction is performed on the operation and the chessboard state respectively to obtain the operation features corresponding to the operation and the chessboard features corresponding to the chessboard state. Based on the operational features and the chessboard features, the object type of the target object is determined, and the intelligent agent corresponding to the object type is determined; The agent simulates the subsequent operations of the target object on the chessboard to obtain the success probability of the current piece generation strategy on the chessboard. Based on the success rate, the piece generation strategy is adjusted to obtain the adjusted piece generation strategy, and the pieces on the chessboard are generated based on the adjusted piece generation strategy.

[0005] This application provides a processing device for a board game, comprising: The acquisition module is used to acquire the operations performed by the target object on the chessboard and the state of the chessboard. An extraction module is used to extract features from the operation and the chessboard state respectively, to obtain the operation features corresponding to the operation and the chessboard features corresponding to the chessboard state. The determination module is used to determine the object type of the target object based on the operation characteristics and the chessboard characteristics, and to determine the intelligent agent corresponding to the object type; An execution module is used to simulate the execution of subsequent operations of the target object on the chessboard through the intelligent agent, and to obtain the success probability of the current piece generation strategy on the chessboard. The generation module is used to adjust the piece generation strategy based on the level completion probability to obtain the adjusted piece generation strategy, and generate the pieces of the chessboard based on the adjusted piece generation strategy.

[0006] In some embodiments, the extraction module is further configured to divide the operation and the chessboard state into units according to a preset time step to obtain the operation and chessboard state corresponding to multiple time steps; and sequentially extract features from the operation and the chessboard state corresponding to each time step to obtain the operation features of the operation and the chessboard features of the chessboard state corresponding to each time step.

[0007] In some embodiments, the determining module is further configured to perform feature filtering on the operation features of the operation corresponding to each time step and the chessboard features of the chessboard state corresponding to each time step, respectively, to obtain the target operation features and target chessboard features corresponding to each time step; according to the order of the time steps, the target operation features and target chessboard features corresponding to each time step are sequentially fused to obtain the first fused feature corresponding to the target operation feature and the second fused feature corresponding to the target chessboard feature; and based on the first fused feature and the second fused feature, the object type of the target object is determined.

[0008] In some embodiments, the determining module is further configured to: set a reward weight for the agent based on the operations performed by the agent on the chessboard; classify the agents based on the actual operations performed by each agent on the chessboard and the reward weight to obtain agents corresponding to each object type; obtain the actual success probability of the target object corresponding to each object type; determine the simulated success probability of the agent corresponding to each object type by having the agent corresponding to each object type operate on the chessboard; and train the agents corresponding to each object type based on the actual success probability and the simulated success probability to obtain trained agents corresponding to each object type.

[0009] In some embodiments, the generation module is further configured to determine the influence weight of each item combination on the success probability, and the conflict coefficient of each item combination on the chessboard; fuse the influence weight and the conflict coefficient to obtain the influence degree of each item combination on the success probability; in response to the success probability being lower than a first threshold, increase the generation probability of item combinations with positive influence degrees to obtain an adjusted piece generation strategy; in response to the success probability being higher than a second threshold, increase the generation probability of item combinations with negative influence degrees to obtain an adjusted piece generation strategy.

[0010] In some embodiments, the generation module is further configured to, for each combination, determine a first probability of passing the level excluding the item combination and a second probability of passing the level including the item combination; determine a first ratio between the second probability of passing the level and the first probability of passing the level; and subtract a preset value from the first ratio to obtain the conflict coefficient of the item combination on the chessboard.

[0011] This application provides an electronic device, the electronic device comprising: Memory is used to store executable instructions or computer programs. The processor, when executing computer-executable instructions or computer programs stored in the memory, implements the board game processing method provided in the embodiments of this application.

[0012] This application provides a computer-readable storage medium storing a computer program or computer-executable instructions for implementing the board game processing method provided in this application when executed by a processor.

[0013] This application provides a computer program product, including a computer program or computer executable instructions. When the computer program or computer executable instructions are executed by a processor, they implement the board game processing method provided in this application.

[0014] The embodiments of this application have the following beneficial effects: The system acquires the operations performed by the target object on the chessboard and the chessboard state. It then extracts features from the operations and the chessboard state to obtain the operation features corresponding to the operations and the chessboard features corresponding to the chessboard state. Based on the operation features and the chessboard features, it determines the object type of the target object and the corresponding agent. By identifying the target object's operations on the chessboard and the chessboard state, the system can determine the object type and thus identify an agent that matches the object type. This ensures type consistency between the agent and the target object and improves the accuracy of subsequent simulations of the agent's operations on the target object.

[0015] Then, an intelligent agent simulates the subsequent operations of the target object on the chessboard to obtain the success probability of the current piece generation strategy. Based on the success probability, the piece generation strategy is adjusted to obtain the adjusted piece generation strategy. Based on the adjusted piece generation strategy, the pieces on the chessboard are generated. By using an intelligent agent of the same object type as the target object, the subsequent operations of the target object are simulated to determine whether the current piece generation strategy is suitable for the current chessboard state. Then, the piece generation strategy is adjusted according to the simulated success probability to generate pieces that conform to the current chessboard state, avoiding situations where the success probability is too low or too high, thus improving the user experience. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the architecture of the board game processing system 100 provided in the embodiments of this application; Figure 2 This is a schematic diagram of the structure of the electronic device 500 provided in the embodiments of this application; Figure 3 This is a flowchart illustrating the board game processing method provided in the embodiments of this application; Figure 4 This is a flowchart illustrating the actual implementation of the board game processing method provided in the embodiments of this application; Figure 5 This is a flowchart of the training agent provided in the embodiments of this application; Figure 6 This is a schematic diagram of the supplementary algorithm provided in the embodiments of this application; Figure 7 This is the processing flow for the special rules provided in the embodiments of this application; Figure 8 This is a schematic diagram of the board game processing system provided in the embodiments of this application.

[0017] It should be noted that the terms "first" and "second" mentioned above are only used to distinguish between different options and do not represent the degree of superiority or inferiority of the options or their priority in the implementation process. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0020] In the following description, the terms "first, second, third" are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first, second, third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0021] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0022] Unless otherwise defined, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in the embodiments of this application is for the purpose of describing the embodiments of this application only and is not intended to limit this application.

[0023] In the implementation of this application, the collection and processing of relevant data should strictly comply with the requirements of relevant laws and regulations, obtain the informed consent or separate consent of the personal information subject, and carry out subsequent data use and processing within the scope of laws and regulations and the authorization of the personal information subject.

[0024] Before providing a further detailed description of the embodiments of this application, the nouns and terms involved in the embodiments of this application will be explained, and the nouns and terms involved in the embodiments of this application shall be interpreted as follows.

[0025] 1) Chessboard: A chessboard is a flat tool used in chess games to carry out the movement, layout and combat of chess pieces. Its core function is to provide clear positional rules for the chess pieces (such as squares and lines) so that the game logic can be executed. A chessboard is usually a flat surface with fixed squares, lines or markings.

[0026] 2) Intelligent Agent: In the field of artificial intelligence, an intelligent agent is a hardware or software entity or system that can perceive its environment through sensors, then make autonomous decisions and take actions based on the environment using actuators to achieve specific goals. Intelligent agents do not require continuous human intervention and can make decisions independently based on built-in algorithms. For example, autonomous vehicles can autonomously plan routes and avoid obstacles based on road conditions without constant driver control. Intelligent agents can perceive environmental changes in real time and respond quickly. For example, a power grid dispatching intelligent agent can respond to fluctuations in power load and adjust power distribution in a very short time. Intelligent agents do not just passively respond to commands, but also actively move towards the goal. For example, a supply chain intelligent agent can predict inventory gaps in advance and proactively trigger replenishment processes. Intelligent agents can continuously optimize their strategies through learning. For example, AlphaGo continuously learns from countless chess games, gradually improving its chess skills and becoming more flexible in its strategies against different opponents.

[0027] This application provides a method, apparatus, electronic device, computer-readable storage medium, and computer program product for processing board games, capable of generating chess pieces that conform to the current board state. The following describes exemplary applications of the electronic device provided in this application. The electronic device provided in this application can be implemented as various types of terminals such as laptops, tablets, desktop computers, set-top boxes, smartphones, smart speakers, smartwatches, smart TVs, and in-vehicle terminals, or it can be implemented as a server. The following will describe exemplary applications when the device is implemented as a server.

[0028] See Figure 1 , Figure 1 This is a schematic diagram of the architecture of the board game processing system 100 provided in the embodiments of this application. In order to support the processing application of a board game, the terminal 400 connects to the server 200 through the network 300. The network 300 can be a wide area network or a local area network, or a combination of the two.

[0029] Terminal 400 is used to acquire the operations performed by the target object on the chessboard and the state of the chessboard, and then transmits the operations performed on the chessboard and the state of the chessboard to server 200 through network 300; server 200 is used to extract features from the operations and the state of the chessboard respectively to obtain the operation features corresponding to the operations and the chessboard features corresponding to the state of the chessboard; based on the operation features and the chessboard features, the object type of the target object is determined, and the agent corresponding to the object type is determined; the agent simulates the subsequent operations of the target object on the chessboard to obtain the success probability of the current piece generation strategy of the chessboard; based on the success probability, the piece generation strategy is adjusted to obtain the adjusted piece generation strategy, and the pieces of the chessboard are generated based on the adjusted piece generation strategy.

[0030] The board game processing method provided in this application embodiment can be applied to various board game scenarios.

[0031] See Figure 2 , Figure 2 This is a schematic diagram of the structure of the electronic device 500 provided in the embodiments of this application, with the electronic device 500 as an example. Figure 1 Taking server 200 as an example, Figure 2 The illustrated electronic device 500 includes at least one processor 510, a memory 550, at least one network interface 520, and a user interface 530. The various components in the electronic device 500 are coupled together via a bus system 540. It is understood that the bus system 540 is used to implement communication between these components. In addition to a data bus, the bus system 540 also includes a power bus, a control bus, and a status signal bus. However, for clarity, ... Figure 2 The general labeled all buses as Bus System 540.

[0032] The processor 510 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0033] User interface 530 includes one or more output devices 531 that enable the presentation of media content, including one or more speakers and / or one or more visual displays. User interface 530 also includes one or more input devices 532, including user interface components that facilitate user input, such as a keyboard, mouse, microphone, touch screen display, camera, other input buttons and controls.

[0034] The memory 550 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state storage, hard disk drives, optical disk drives, etc. The memory 550 may optionally include one or more storage devices physically located away from the processor 510.

[0035] The memory 550 may include volatile memory or non-volatile memory, or both. The non-volatile memory may be read-only memory (ROM), and the volatile memory may be random access memory (RAM). The memory 550 described in this application embodiment is intended to include any suitable type of memory.

[0036] In some embodiments, memory 550 is capable of storing data to support various operations, examples of which include programs, modules, and data structures or subsets or supersets thereof, as illustrated below.

[0037] Operating system 551 includes system programs for handling various basic system services and performing hardware-related tasks, such as the framework layer, core library layer, driver layer, etc., for implementing various basic business functions and handling hardware-based tasks; The network communication module 552 is used to reach other electronic devices via one or more (wired or wireless) network interfaces 520, exemplary network interfaces 520 including: Bluetooth, WiFi, and Universal Serial Bus (USB), etc. Presentation module 553 is used to enable the presentation of information (e.g., user interface for operating peripheral devices and displaying content and information) via one or more output devices 531 (e.g., display screen, speaker, etc.) associated with user interface 530. The input processing module 554 is used to detect and translate one or more user inputs or interactions from one or more input devices 532.

[0038] In some embodiments, the apparatus provided in this application can be implemented in software. Figure 2 A processing device 555 for a board game stored in memory 550 is shown. This device can be software in the form of programs and plug-ins, and includes the following software modules: an acquisition module 5551, an extraction module 5552, a determination module 5553, an execution module 5554, and a generation module 5555. These modules are logically connected and can therefore be arbitrarily combined or further separated according to their implemented functions. The functions of each module will be described below.

[0039] In other embodiments, the apparatus provided in this application can be implemented in hardware. As an example, the apparatus provided in this application can be a processor in the form of a hardware decoding processor, which is programmed to execute the board game processing method provided in this application. For example, the processor in the form of a hardware decoding processor can be one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0040] The following describes the board game processing method provided in the embodiments of this application. As mentioned above, the electronic device implementing the board game processing method of the embodiments of this application can be a server, a terminal, or a combination of both. Therefore, the executing entity of each step will not be described again below.

[0041] See Figure 3 , Figure 3 This is a first flowchart illustrating the board game processing method provided in this application embodiment, which will be combined with... Figure 3 The steps shown are explained.

[0042] In step 101, the operations performed by the target object on the chessboard and the state of the chessboard are obtained.

[0043] As an example, the target object refers to the object that operates on the chessboard, i.e., the user running the chessboard game. The operations performed by the target object on the chessboard can be the steps that the target object has already performed during the chessboard game. Taking a match-3 game (the core gameplay is to exchange adjacent game elements (such as gems, fruits, blocks, etc.) to connect 3 or more identical elements horizontally or vertically to eliminate these elements and score points, while triggering chain reactions, item generation, and other additional gameplay) as an example, the operations performed by the target object on the chessboard can be the operations of exchanging pieces. The state of the chessboard can include the piece information of each position on the chessboard after each piece exchange by the user. It should be noted that since the operations performed by the target object on the chessboard are directly related to the order of operations, the execution order of each operation needs to be recorded when obtaining the operations performed by the target object on the chessboard.

[0044] In step 102, feature extraction is performed on the operation and the chessboard state respectively to obtain the operation features corresponding to the operation and the chessboard features corresponding to the chessboard state.

[0045] In some embodiments, the feature extraction of the operation and the chessboard state in step 102 to obtain the operation features corresponding to the operation and the chessboard features corresponding to the chessboard state can be achieved by the following technical solution: the operation and the chessboard state are divided into units according to a preset time step to obtain the operation and chessboard state corresponding to multiple time steps; the operation and the chessboard state corresponding to each time step are extracted sequentially to obtain the operation features of the operation corresponding to each time step and the chessboard features of the chessboard state corresponding to each time step.

[0046] As an example, a time step can refer to a preset unit used to divide a continuous sequence of operations and a sequence of chessboard states into discrete units. A time step can be defined based on the number of operations (e.g., every 5 operations) or the length of time (e.g., every 10 seconds).

[0047] As an example, dividing the executed operations and chessboard states into units based on time steps can be done by dividing the multiple executed operations and chessboard states generated by the target object according to the time span or the number of operation steps. For example, if the target object has executed 20 operations, and the time step is set to 10, then the first 10 operations executed by the target object and the corresponding chessboard states can be used as the operations and chessboard states corresponding to the first time step. If the total time taken for the operations executed by the target object is 3 minutes, then the operations executed by the target object in the first minute and the corresponding chessboard states can be used as the operations and chessboard states corresponding to the first time step, the operations executed by the target object in the second minute and the corresponding chessboard states corresponding to the second time step, and the operations executed by the target object in the third minute and the corresponding chessboard states corresponding to the third time step.

[0048] As an example, operational features are vectorized data that characterize the operational habits and strategic tendencies of a target object over a period of time. For example, they may include the probability of consecutive eliminations, the proportion of invalid operations, and the item preference index. Chessboard features are spatial quantitative data that characterize the layout of the chessboard and potential game trends, including regional value density, connectivity topology, and symmetry defects.

[0049] As an example, for the operations performed by the target object at each time step, the operations performed by the target object at each time step can be input into a pre-trained feature extraction network. Features such as the probability of chain elimination, the proportion of invalid operations, and the item preference index can be extracted. In addition, the effective operation decay rate of the target object within 10 consecutive steps can be calculated, the timing of the target object's attention decline can be predicted, the number of chain eliminations within 5 steps after the use of the gain piece can be recorded, and the benefits of item combination (such as bomb + rainbow candy) can be quantified.

[0050] As an example, the feature extraction network can be a Convolutional Neural Network (CNN), which is a core model in deep learning specifically for processing grid-structured data (such as images, videos, and audio). It is most widely used in the field of computer vision. CNN is a type of deep feedforward neural network that includes convolutional computation. Through three major mechanisms—local connectivity, weight sharing, and pooling—it can efficiently extract spatial and temporal features and reduce the number of parameters, avoiding the curse of dimensionality of fully connected networks.

[0051] The feature extraction network extracts features from the chessboard state at each time step. The chessboard state can be represented as a matrix, where the numbers represent: 0 for an empty space, 1 for a red candy, 2 for a blue candy, 3 for an obstacle, and 4 for a bomb. The corresponding matrix for the chessboard can then be: [1, 2, 0, 3, 1] [3, 4, 2, 1, 0] [2, 1, 3, 4, 2] The corresponding matrix of the chessboard can then be input into a pre-trained feature extraction network. The feature extraction network extracts chessboard features such as regional value density, connectivity topology, and symmetry defects. Among them, regional value density refers to the proportion of high-value elements (such as bombs) in a 3×3 local area, which affects the player's decision priority. Connectivity topology identifies isolated areas (such as empty spaces surrounded by obstacles) through dilated convolution and predicts the risk of blocking movement paths. Symmetry defects are the symmetry imbalance of the chessboard layout (such as the bias of target elements), which leads to a sharp increase in the difficulty of operation.

[0052] As an example, the specific process of a feature extraction network extracting the operations performed on the chessboard and the state of the chessboard can be as follows: the game chessboard is represented as a numerical matrix (e.g., 0=space, 1=red candy, 2=ice block obstacle, 3=special effect marker), and then the matrix is ​​converted into a four-dimensional tensor [batch size, number of channels, height, width], where the number of channels = 3, representing the basic element layer, obstacle layer, and special effect layer, respectively. Then, the feature values ​​are mapped to the [0, 1] interval through normalization.

[0053] During the extraction process, 16 3×3 convolutional kernels can be used to traverse the chessboard. Each convolutional kernel detects specific local patterns, such as: potential areas of horizontal / vertical triplet (e.g., adjacent candies of the same color); areas of clustered obstacles (e.g., areas of dense ice blocks); and special effect trigger points (e.g., bomb generation locations). Finally, the extracted features are non-linearly activated. For example, the extracted features can be input into an activation function (ReLU function) to obtain the activated feature map.

[0054] Finally, 2×2 pooling can be performed on the 9×9 feature map to output a 4×4 feature map. This operation retains the maximum value of each 2×2 window (such as the highest elimination priority within the region), filters out low-relevance information, and then flattens the 4×4×16 feature map into a 256-dimensional vector, which serves as the obtained operational features and chessboard features.

[0055] The above methods can accurately extract operational features and chessboard features.

[0056] In step 103, based on the operation features and the chessboard features, the object type of the target object is determined, and the intelligent agent corresponding to the object type is determined.

[0057] In some embodiments, the operation features include operation features corresponding to multiple time steps, and the chessboard features include chessboard features of chessboard states corresponding to multiple time steps; the determination of the object type of the target object based on the operation features and the chessboard features in step 103 can be achieved by the following technical solution: performing feature filtering on the operation features corresponding to each time step and the chessboard features of the chessboard states corresponding to each time step to obtain the target operation features and target chessboard features corresponding to each time step; sequentially fusing the target operation features and target chessboard features corresponding to each time step according to the order of the time steps to obtain the first fused feature corresponding to the target operation feature and the second fused feature corresponding to the target chessboard feature; and determining the object type of the target object based on the first fused feature and the second fused feature.

[0058] As an example, this application can use a Long Short-Term Memory (LSTM) network to determine the object type of the target object. The core of LSTM is to introduce cell state and three gate control mechanisms, so that the model can autonomously control the retention, updating and output of memory.

[0059] As an example, feature filtering is performed on the operation features and chessboard features corresponding to each time step to obtain the target operation features and target chessboard features corresponding to each time step. This involves removing invalid operations and useless information from the operation features and chessboard features. For example, features representing terminal information and network information of the target object are included in the operation features, and features representing the chessboard pattern are included in the chessboard features.

[0060] Specifically, the forgetting gate method can be used to filter operational features and chessboard features, as shown in the following formula (1): (1) In formula (1), It is the output vector at time t. It is an activation function. This is the weight matrix. It is the hidden state vector at time (t-1). It is the input vector at time t. right and Perform feature splicing. It is the bias vector.

[0061] in, The value range is ([0,1]) (constrained by the Sigmoid activation function). Element-wise multiplication using the Hadamard product achieves dimensional filtering of cell states. The closer a certain dimension value is to 1, the better. The more complete the corresponding dimension information, the more it is retained; the closer it is to 0, the more the corresponding dimension information is filtered out (reset to zero). For example, if... = [0.9, 0.1, 0.8]), If the value is [10, 20, 30], then after filtering... = [9, 2, 24] (The second dimension information is heavily filtered).

[0062] As an example, the target operation features and target chessboard features corresponding to each time step can be fused through the input gate. For details, please refer to the following formula (2): (2) In formula (2), It is the input vector at time t. It is the output of the forget gate at time t. It is the element-wise multiplication of two vectors of the same dimension. yes The input vector at time t, It is the output weight vector of the input gate. It is the information to be memorized input at time t.

[0063] Using the above formula, the features corresponding to each time step are fused and input sequentially, and then the output vector is adjusted according to the following formula (3).

[0064] (3) In formula (3), The output vector, It is an activation function. This is the weight matrix of the output gate. It is the hidden state vector at time (t-1). It is the input vector at time t. right and Perform feature splicing. It is the bias vector of the output gate.

[0065] Using the above method, the target operation features can be sequentially input into the forget gate, input gate, and output gate to obtain the first fusion feature. Similarly, the target chessboard features can be sequentially input into the forget gate, input code, and output gate to obtain the second fusion feature. Finally, the obtained first and second fusion features can be input into the fully connected layer, which outputs the predicted probability of each object type. For example, if the object types include novice, intermediate, and hardcore types, and the probabilities output by the fully connected layer are 0.1, 0.7, and 0.2 respectively, then the object type of the target object can be determined to be the intermediate type.

[0066] By combining the operations performed on the target object, the object type of the target object can be accurately identified.

[0067] In some embodiments, before determining the agent corresponding to the object type in step 103, the following technical solution may also be performed: based on the operations performed by the agent on the chessboard, a reward weight for the agent is set, and based on the actual operations performed by each agent on the chessboard and the reward weight, the agents are classified to obtain agents corresponding to each object type; the actual success probability of the target object corresponding to each object type is obtained; the agent corresponding to each object type operates on the chessboard to determine the simulated success probability of the agent corresponding to each object type; based on the actual success probability and the simulated success probability, the agent corresponding to each object type is trained to obtain the trained agent corresponding to each object type.

[0068] As an example, the probability of passing a level can refer to a quantitative prediction of successful level passing obtained by an agent that matches the type of the target object, under a given board state and current piece generation strategy, by simulating the execution of subsequent operations until the end of the game (success or failure) through multiple rehearsals. It is usually expressed as a probability value or percentage between 0 and 1.

[0069] As an example, a reward function can be set as shown in the following formula (4): (4) In formula (4), It is the score of the intelligent agent. , and As a reward weight, and These correspond to the agent's positive actions on the chessboard. Taking a match-3 game as an example... It can perform consecutive elimination operations. It can be used to determine the degree of goal completion. It allows for negative actions by the AI ​​agent on the chessboard; for example, in a match-3 game. This could be an invalid operation.

[0070] As an example, multiple agents play multiple games, and the score of each agent is obtained by formula (4) above. The agents are classified according to their scores to obtain agents corresponding to each object type. For example, agents with a score of no more than 30 points are the agents corresponding to the novice type, agents with a score of more than 30 points but no more than 70 points are the agents corresponding to the advanced type, and agents with a score of more than 70 points are the agents corresponding to the hardcore type.

[0071] After obtaining the agent corresponding to each type, the agent of each type can be further trained. First, by statistics, the actual success probability of the target object corresponding to each object type is determined. For example, the actual success probability of the target object corresponding to the novice type is 50%, the actual success probability of the target object corresponding to the intermediate type is 70%, and the actual success probability of the target object corresponding to the hardcore type is 90%.

[0072] For each object type, the execution strategy of the agent is adjusted with the goal of minimizing the difference between the actual success probability of the target object and the simulated success probability of the agent. Training of the agent is stopped when the difference between the actual success probability of the target object and the simulated success probability of the agent meets the preset difference threshold.

[0073] The above methods can improve the accuracy of intelligent agents simulating target objects to perform operations.

[0074] In step 104, the agent simulates the subsequent operations of the target object on the chessboard to obtain the success probability of the current piece generation strategy on the chessboard.

[0075] As an example, a piece generation strategy can refer to an algorithm or set of rules that determines the type, position, and probability of new pieces being generated on the board. A piece generation strategy includes at least generation probability parameters for different combinations of items, and the probability parameters can be dynamically adjusted according to the probability of passing the level.

[0076] As an example, by having an agent simulate the target object's subsequent operations under the current chessboard state, the probability of passing the game based on the simulated chessboard state can be determined. This probability of passing the game represents the predicted probability of passing the target object without making any changes.

[0077] In step 105, the piece generation strategy is adjusted based on the level completion probability to obtain an adjusted piece generation strategy, and the pieces on the chessboard are generated based on the adjusted piece generation strategy.

[0078] In some embodiments, adjusting the piece generation strategy based on the level completion probability in step 105 to obtain an adjusted piece generation strategy, and generating pieces for the chessboard based on the adjusted piece generation strategy, can be achieved through the following technical solution: determining the influence weight of each item combination on the level completion probability, and the conflict coefficient of each item combination on the chessboard; fusing the influence weight and the conflict coefficient to obtain the influence degree of each item combination on the level completion probability; increasing the generation probability of item combinations with positive influence degrees in response to the level completion probability being lower than a first threshold, thereby obtaining the adjusted piece generation strategy; and increasing the generation probability of item combinations with negative influence degrees in response to the level completion probability being higher than a second threshold, thereby obtaining the adjusted piece generation strategy.

[0079] As an example, the conflict coefficient of different item combinations reflects the impact of the item combination on the probability of passing the level on any board. For example, the conflict coefficient of an item combination that requires three eliminations in a position is less than that of an item combination that requires four eliminations in a position. In addition, the conflict coefficient of an item combination can be positive or negative. When the conflict coefficient of an item combination is positive, it means that the item combination can guarantee the target object to pass the level and increase the target object's probability of passing the level. When the conflict coefficient of an item combination is negative, it means that the item combination will hinder the target object from passing the level and reduce the target object's probability of passing the level.

[0080] The influence weight of item combinations represents their impact on the probability of clearing the game in the current board. For example, there may be an item combination that requires at least three piece exchanges to eliminate. When the number of exchanges is less than five, the influence weight of this item combination is higher than the influence weight when the number of exchanges is twenty.

[0081] As an example, the influence degree of the prop combination obtained by fusing the conflict coefficient and the influence weight directly reflects the degree of influence of the prop combination on the current board. The method of fusing the conflict coefficient and the influence weight can directly multiply the influence weight and the conflict coefficient, and use the product result as the influence degree of the prop combination.

[0082] As an example, when the probability of passing the level is lower than the first threshold, it indicates that the predicted target object will have difficulty passing the level under the current piece generation strategy. In order to ensure the experience of the target object, the generation probability of the prop combination with a positive influence value can be increased, that is, the generation probability of the prop combination that helps the target object pass the level can be increased.

[0083] As an example, when the success rate is higher than the second threshold, it indicates that the target object is likely to easily complete the level. In order to improve the target object's experience and ensure the difficulty of the level, the generation probability of item combinations with negative influence can be increased, that is, the generation probability of item combinations that hinder the target object from completing the level can be increased.

[0084] The above method allows for dynamic adjustment of the piece generation strategy based on the predicted success rate, ensuring a positive user experience for the target audience.

[0085] In some embodiments, the determination of the conflict coefficient of each item combination on the chessboard can be achieved by the following technical solution: for each combination, a first probability of passing the level excluding the item combination and a second probability of passing the level including the item combination are determined; a first ratio of the second probability of passing the level to the first probability of passing the level is determined, and the difference between the preset value and the first ratio is calculated to obtain the conflict coefficient of the item combination on the chessboard.

[0086] As an example, the method for determining the conflict coefficient of a prop combination can be found in the following formula (5): (5) As an example, in formula (5), This represents the conflict coefficient of the item combination; 1 is the default value. It is the probability of clearing the first level. It's the second probability of clearing the level. The second ratio is calculated by the agent for each item combination. The agent simulates the first probability of passing the level without the item combination and the second probability of passing the level without the item combination. Then, the first ratio of the first probability of passing the level and the second probability of passing the level is determined. Finally, the conflict coefficient of the item combination is obtained by subtracting the first ratio from 1.

[0087] As an example, if the expected success rate of a certain level is 70% when there is no conflict, the actual success rate drops to 42% after adding the "teleportation portal + spikes" combination. Then the conflict coefficient is 1 - (42 / 70) = 0.4. In addition, other factors can be superimposed. For example, if a 3×3 scan finds that the failure rate of a certain area is >60%, the conflict coefficient of such "high death area" will increase by an additional 0.2.

[0088] The following is combined Figure 4 This will illustrate an exemplary application of the embodiments of this application in a real-world application scenario. Figure 4 This is a flowchart illustrating the actual implementation of the board game processing method provided in this application.

[0089] Object modeling module: The data acquisition layer can capture the operation sequence of objects in real time (coordinates before and after swapping, number of steps when using items, frequency of chain elimination triggers), and divide the time window into 5 steps / units.

[0090] For the operation data within each time window, construct a feature vector: V = [probability of consecutive eliminations, percentage of invalid operations, item preference index], normalize it, and input it into the model.

[0091] The chessboard state matrix is ​​represented as follows: 0 = empty space, 1 = red candy, 2 = blue candy, 3 = obstacle, 4 = bomb. [1, 2, 0, 3, 1] [3, 4, 2, 1, 0] [2, 1, 3, 4, 2] In addition to obstacle distribution, feature extraction networks can also extract the following spatial features: Regional value density: Calculates the proportion of high-value elements (such as bombs) within a 3×3 local area, which affects the player's decision priority.

[0092] Connectivity topology: Identify isolated regions (such as empty spaces surrounded by obstacles) through dilated convolutions to predict the risk of movement path congestion.

[0093] Symmetry defects: Detecting symmetry imbalances in the chessboard layout (such as target element offset) significantly increases the difficulty of operation.

[0094] Operational fatigue cycle: Statistically analyze the rate of effective operation decay of players within 10 consecutive steps to predict when attention will decline.

[0095] Chain Reaction of Items: Records the number of chain eliminations within 5 steps after an item is used, and quantifies the benefits of item combinations (such as bomb + rainbow candy). Specifically, the 9×9 game board can be represented as a numerical matrix (e.g., 0 = empty space, 1 = red candy, 2 = ice block obstacle, 3 = special effect marker); the matrix is ​​converted into a 4D tensor [batch size, number of channels, height, width], where the number of channels = 3 (basic element layer, obstacle layer, special effect layer), and the values ​​are mapped to the [0,1] interval by normalization to improve training stability.

[0096] Kernel sliding scan: 16 3×3 convolutional kernels (kernel_size=3, padding=1) are used to traverse the chessboard. Each convolutional kernel detects specific local patterns, such as: potential horizontal / vertical triplet regions (e.g., adjacent candies of the same color); obstacle clusters (e.g., dense ice blocks); special effect trigger points (e.g., bomb generation locations); then non-linear activation is performed, activating the feature map through the ReLU function max(0,x), preserving salient features (e.g., high-density eliminable regions) and suppressing noise (e.g., isolated spaces).

[0097] Perform 2×2 pooling (stride=2) on the 9×9 feature map to output a 4×4 feature map. This operation retains the maximum value of each 2×2 window (such as the highest elimination priority within the region), filters out low-relevance information, and flattens the 4×4×16 feature map into a 256-dimensional vector, which is used as the input unit of the temporal network, completing the transformation from spatial structure to high information density vector.

[0098] In the temporal network, the player's operation sequence (such as exchanging coordinates, using items) is divided into 5 steps / units. Each time window is associated with a 256-dimensional feature vector output by a feature extraction network (i.e., the operation features and chessboard features mentioned above). Then, the 256-dimensional feature vector is input into the forget gate to determine how many historical states are retained (such as discarding early invalid operation patterns). The specific processing logic of the forget gate can be shown in the following formula (6): (6) In formula (6), It is the output vector at time t. It is an activation function. This is the weight matrix. It is the hidden state vector at time (t-1). It is the input vector at time t. right and Perform feature splicing. It is the bias vector.

[0099] The features retained by the forget gate are then input into the input gate. The processing logic of the input gate is shown in the following formula (7): (7) In formula (7), It is the input vector at time t. It is the output of the forget gate at time t. It is the element-wise multiplication of two vectors of the same dimension. yes The input vector at time t, It is the output weight vector of the input gate. It is the information to be memorized input at time t.

[0100] Finally, the features processed by the input gate are input into the output gate. The processing logic of the output gate is shown in the following formula (8): (8) In formula (8), The output vector, It is an activation function. This is the weight matrix of the output gate. It is the hidden state vector at time (t-1). It is the input vector at time t. right and Perform feature splicing. It is the bias vector of the output gate.

[0101] Finally, the hidden state features at the last time step are input into the fully connected layer, and the output consists of three types of probabilities: 0 = Novice (operation interval > 2 seconds, high percentage of invalid operations) 1 = Advanced (Medium combo frequency, occasional use of items) 2 = Hardcore (predicts 3 or more steps, frequently triggers item combinations).

[0102] Temporal networks can avoid gradient vanishing through linear propagation of cell states and capture long-term strategies of more than 50 steps (such as a novice suddenly mastering the skill of combining items in the later stages).

[0103] exist Figure 4In the level simulation engine, three types of Artificial Intelligence (AI) players (i.e., the aforementioned intelligent agents) can be trained in different roles, simulating the behavioral patterns of the three types of players (e.g., beginners tend to click randomly, while hardcore players plan combos). Each AI has its own learning strategy, but they share a common set of reward rules: points are awarded for consecutively eliminating 3 or more elements (×1.2 weight); points are awarded for each target element eliminated (×0.8 weight); points are deducted for clicking without eliminating any elements (×0.5 weight); and game levels are simulated, allowing the three types of AI to be trained simultaneously.

[0104] The training process is as follows Figure 5 As shown, Figure 5 This is a flowchart of the training agent provided in the embodiments of this application.

[0105] exist Figure 5 In this game, AI teams attempt levels in parallel, with three different AI types each attempting the same level 1000 times. Parameters during the level completion process are analyzed, such as: Level completion rate = (Number of successful attempts ÷ 1000) × 100% (e.g., 38% indicates a relatively difficult level); Average number of attempts = (Number of failures for beginners + Number of failures for advanced players + Number of failures for hardcore players) ÷ 3 (reflecting the average player's frustration); Step utilization rate = Actual number of steps consumed ÷ Level's step limit × 100% (>60% indicates a reasonable pace).

[0106] Next, scan the board and mark areas with a failure rate >60% (such as the dense ice block area in the lower right corner). If the hardcore player's success rate is <40%, automatically increase the probability of bomb generation (+15%). If the move utilization rate is <50%, use GAN to generate a new layout with fewer obstacles.

[0107] Next, the element conflict matrix is ​​determined. Specifically, the influence weight represents the magnitude of the influence of different conflict types on the completion rate. The influence weight is a "problem severity score". The larger the value, the stronger the obstacle that this type of conflict poses to the player's completion of the level. The conflict coefficient of the item combination measures the actual severity of the current conflict, ranging from 0 to 1. Its calculation is based on actual test data. Specifically, the expected completion rate without conflict can be measured first, and then compared with the actual completion rate after adding conflict. For example, if the expected completion rate of a certain level is 70% without conflict, and the actual completion rate drops to 42% after adding the "teleportation portal + spike" combination, then the conflict coefficient is 1 - (42 / 70) = 0.4. In addition, other factors can be superimposed. For example, if a 3×3 scan finds that the failure rate of a certain area is >60%, the conflict coefficient of this type of "high death area" will be increased by an additional 0.2.

[0108] If the scan finds a 72% failure rate in the "icing + chocolate" area in the lower right corner, it is marked as a high-risk conflict area, with a weight Wi=0.18 and a conflict coefficient c. i When the value is 0.7, the influence of the item combination is ΔR=0.126 (the success rate decreases by 12.6%). In this case, the probability of chocolate generation can be reduced from 25% to 15%; the frequency of the appearance of the "Rainbow Candy" clearing item can be increased.

[0109] Figure 4 The dynamic optimization closed-loop module in the middle: It can monitor failure reason codes in real time (such as "target element not eliminated 10 times consecutively"), triggering compensation algorithms, such as... Figure 6 As shown, Figure 6 This is a schematic diagram of the supplementary algorithm provided in the embodiments of this application.

[0110] exist Figure 6 In this process, the target object can be compensated after failure by adjusting the generation probability of the target element and the method of generating a new chessboard layout. The method of regenerating the chessboard is to input a 64×64 heatmap (marking high failure areas) and the object encoding vector of the target object into the generator to generate a corresponding new chessboard layout. The discriminator then determines whether the generated new chessboard layout is reasonable based on the feature library of historical successful levels.

[0111] Figure 4 The special rule processing engine in the system can perform actions such as Figure 7 The steps shown, Figure 7 This is the processing flow for the special rules provided in the embodiments of this application.

[0112] like Figure 7 As shown, item combinations with excessively high conflict coefficients can significantly impact the target player's chances of completing the game, causing them to overlook the overall game and focus solely on the high-conflict item combinations, thus reducing the user experience. Therefore, item combinations with excessively high conflict coefficients can be modified and supplemented. Specifically, the impact of item combinations on the normal game board can be mitigated by adding a compensation term to the reward function. For example, for the item combination of bomb and ice, the explosion range can be reduced to the original value x 0.8, and for the item combination of portal and poison gas, the probability of contaminating the displacement path can be increased to the original value x 1.2.

[0113] To more clearly illustrate the collaborative working relationships and data flow paths among the core components of this invention, please refer to [link / reference needed]. Figure 8 , Figure 8 This is a schematic diagram of the board game processing system provided in this application embodiment. Logically, the system architecture can be divided into three main parts: data acquisition, core processing, and strategy application, which together form a dynamic and adaptive closed loop. In one embodiment, such as Figure 8 As shown, the system first monitors the chessboard game environment in real time (A1) in section A. Game Environment and Data Acquisition, and captures the operation sequence of the target object (e.g., the player) and the chessboard state data (A2). This data is the basis for all subsequent analyses, containing the player's continuous operation behavior and the layout information of the chessboard at various points in time. Next, the collected data is sent to the B. core processing system. Specifically: The data first flows into the object modeling module (B1). This module is responsible for deep feature extraction (e.g., through the CNN and LSTM networks mentioned in the previous embodiments) of the input operation sequence and board state data. Its core objective is to identify and output the player's current object type. This object type is a precise profile of the player's game level and behavior patterns, such as being categorized as "beginner," "intermediate," or "hardcore."

[0114] Meanwhile, the agent training module (B2), as a key supporting module, is trained either offline or online. Based on a large amount of game data and a preset reward function, this module trains agents that simulate the behavior patterns of different object types, forming a "trained agent pool" containing various types of agents. The level simulation engine (B3) receives the "object type" output from the object modeling module (B1). Based on this type, the engine selects a matching agent from the agent pool provided by the agent training module (B2) (e.g., if the player is identified as a "novice," the "novice agent" is invoked). Subsequently, this agent simulates and executes multiple subsequent moves in the current board state, thereby proactively predicting the probability of this type of player completing the level under the given piece generation strategy. The predicted "probability of success" is passed to the dynamic optimization closed-loop module (B4). This module is the core of the system's decision-making process; it compares the received probability of success with preset thresholds (e.g., the first and second thresholds mentioned in claim 5). Based on the comparison results, the module calculates a specific adjustment scheme for the current piece generation strategy, such as increasing or decreasing the generation probability of specific item combinations. Finally, in section C, Strategy Application, the decision results of the dynamic optimization closed-loop module (B4) are output as an adjusted piece generation strategy (C1). This new strategy is directly applied back to the ongoing board game environment (A1), thereby influencing the generation of subsequent pieces in real time and dynamically. It should be noted that, through Figure 8The architecture shown in this application implements a complete adaptive closed loop, from deep recognition of player behavior to forward-looking probabilistic prediction based on matching agents, and then to fine-tuning of strategies based on the prediction results. The various modules of this architecture work together, enabling the piece generation strategy to no longer be static or simply reactive, but to proactively adapt to different types of players and dynamically optimize the game experience.

[0115] The above methods enable precise control over the difficulty of board games, improve players' flow experience, enhance dynamic optimization capabilities, allow for timely adjustments, improve the ability to handle special rules, and reduce the manpower required for high-quality game levels.

[0116] The following description continues to illustrate the exemplary structure of the board game processing device 555 provided in the embodiments of this application as a software module. In some embodiments, such as... Figure 2 As shown, the software module in the board game processing device 555 stored in the memory 550 may include: The acquisition module 5551 is used to acquire the operations performed by the target object on the chessboard and the state of the chessboard. Extraction module 5552 is used to extract features from the operation and the chessboard state respectively, to obtain the operation features corresponding to the operation and the chessboard features corresponding to the chessboard state; The determination module 5553 is used to determine the object type of the target object based on the operation features and the chessboard features, and to determine the intelligent agent corresponding to the object type; The execution module 5554 is used to simulate the subsequent operations of the target object on the chessboard through the intelligent agent, and obtain the success probability of the current piece generation strategy of the chessboard. The generation module 5555 is used to adjust the piece generation strategy based on the level completion probability to obtain the adjusted piece generation strategy, and generate the pieces of the chessboard based on the adjusted piece generation strategy.

[0117] In some embodiments, the extraction module 5552 is further configured to divide the operation and the chessboard state into units according to a preset time step to obtain the operation and chessboard state corresponding to multiple time steps; and sequentially extract features from the operation and the chessboard state corresponding to each time step to obtain the operation features of the operation and the chessboard features of the chessboard state corresponding to each time step.

[0118] In some embodiments, the determining module 5553 is further configured to perform feature filtering on the operation features of the operation corresponding to each time step and the chessboard features of the chessboard state corresponding to each time step, respectively, to obtain the target operation features and target chessboard features corresponding to each time step; according to the order of the time steps, the target operation features and target chessboard features corresponding to each time step are sequentially fused to obtain the first fused feature corresponding to the target operation feature and the second fused feature corresponding to the target chessboard feature; and based on the first fused feature and the second fused feature, the object type of the target object is determined.

[0119] In some embodiments, the determining module 5553 is further configured to: set the reward weight of the agent based on the operation performed by the agent on the chessboard; classify the agents based on the actual operation performed by each agent on the chessboard and the reward weight to obtain agents corresponding to each object type; obtain the actual success probability of the target object corresponding to each object type; determine the simulated success probability of the agent corresponding to each object type by having the agent corresponding to each object type operate on the chessboard; and train the agents corresponding to each object type based on the actual success probability and the simulated success probability to obtain trained agents corresponding to each object type.

[0120] In some embodiments, the generation module 5555 is further configured to determine the influence weight of each item combination on the success probability, and the conflict coefficient of each item combination on the chessboard; fuse the influence weight and the conflict coefficient to obtain the influence degree of each item combination on the success probability; in response to the success probability being lower than a first threshold, increase the generation probability of item combinations with positive influence degrees to obtain an adjusted piece generation strategy; in response to the success probability being higher than a second threshold, increase the generation probability of item combinations with negative influence degrees to obtain an adjusted piece generation strategy.

[0121] In some embodiments, the generation module 5555 is further configured to, for each combination, determine a first probability of passing the level excluding the item combination and a second probability of passing the level including the item combination; determine a first ratio between the second probability of passing the level and the first probability of passing the level; and subtract a preset value from the first ratio to obtain the conflict coefficient of the item combination on the chessboard.

[0122] This application provides a computer program product comprising a computer program or computer-executable instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer-executable instructions from the computer-readable storage medium and executes the computer-executable instructions, causing the electronic device to perform the board game processing method described above in this application.

[0123] This application provides a computer-readable storage medium storing computer-executable instructions or a computer program. When the computer-executable instructions or the computer program are executed by a processor, the processor will execute the board game processing method provided in this application embodiment. For example, ... Figure 3 The method for handling board games is shown.

[0124] In some embodiments, the computer-readable storage medium may be a memory such as RAM, ROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or it may be a variety of devices including one or any combination of the above-mentioned memories.

[0125] In some embodiments, computer-executable instructions may take the form of programs, software, software modules, scripts, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as stand-alone programs or as modules, components, subroutines, or other units suitable for use in a computing environment.

[0126] As an example, computer-executable instructions may, but do not necessarily, correspond to files in a file system. They may be stored as part of a file that holds other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple co-located files (e.g., files that store one or more modules, subroutines, or code sections).

[0127] As an example, computer-executable instructions can be deployed to execute on a single electronic device, or on multiple electronic devices located at one location, or on multiple electronic devices distributed across multiple locations and interconnected via a communication network.

[0128] In summary, the embodiments of this application can achieve the following beneficial effects: The system acquires the operations performed by the target object on the chessboard and the chessboard state. It then extracts features from the operations and the chessboard state to obtain the operation features corresponding to the operations and the chessboard features corresponding to the chessboard state. Based on the operation features and the chessboard features, it determines the object type of the target object and the corresponding agent. By identifying the target object's operations on the chessboard and the chessboard state, the system can determine the object type and thus identify an agent that matches the object type. This ensures type consistency between the agent and the target object and improves the accuracy of subsequent simulations of the agent's operations on the target object.

[0129] Then, an intelligent agent simulates the subsequent operations of the target object on the chessboard to obtain the success probability of the current piece generation strategy. Based on the success probability, the piece generation strategy is adjusted to obtain the adjusted piece generation strategy. Based on the adjusted piece generation strategy, the pieces on the chessboard are generated. By using an intelligent agent of the same object type as the target object, the subsequent operations of the target object are simulated to determine whether the current piece generation strategy is suitable for the current chessboard state. Then, the piece generation strategy is adjusted according to the simulated success probability to generate pieces that conform to the current chessboard state, avoiding situations where the success probability is too low or too high, thus improving the user experience.

[0130] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, and improvements made within the spirit and scope of this application are included within the scope of protection of this application.

Claims

1. A method for processing a board game, characterized in that, The method includes: Obtain the operations performed by the target object on the chessboard and the state of the chessboard; Feature extraction is performed on the operation and the chessboard state respectively to obtain the operation features corresponding to the operation and the chessboard features corresponding to the chessboard state. Based on the operational features and the chessboard features, the object type of the target object is determined, and the intelligent agent corresponding to the object type is determined; The agent simulates the subsequent operations of the target object on the chessboard to obtain the success probability of the current piece generation strategy of the chessboard, and counts the failure results and step utilization rate of the agent's simulated execution of the subsequent operations. The step utilization rate is the ratio of the number of steps actually consumed by the agent in simulating the execution of the subsequent operations to the maximum number of steps of the chessboard. Based on the failure results, the chessboard is scanned by a scanning window of a preset size, and the areas of the chessboard with a failure rate exceeding a preset failure rate threshold are marked as target areas. Determine the weight of each item combination on the probability of passing the level; For each combination, determine a first probability of clearing the level excluding the item combination and a second probability of clearing the level including the item combination; Determine the first ratio between the second success probability and the first success probability, and subtract the preset value from the first ratio to obtain the conflict coefficient of the item combination on the chessboard; In response to the fact that the target area includes a piece corresponding to the item combination, a preset correction value is added to the conflict coefficient of the item combination to obtain the corrected conflict coefficient of the item combination; The influence weights and the corrected conflict coefficients are fused to obtain the degree of influence of each item combination on the probability of passing the level. In response to the probability of passing the level being lower than a first threshold, the generation probability of the item combination with a positive influence value is increased to obtain an adjusted piece generation strategy; In response to the probability of passing the level being higher than the second threshold, the generation probability of the item combination with a negative influence value is increased to obtain the adjusted piece generation strategy; In response to the step utilization rate being lower than a third threshold, the object encoding vector of the target object is obtained, and the heat map marked with the target region and the object encoding vector of the target object are input into the generator of the generative adversarial network, and a new chessboard layout corresponding to the chessboard is generated through the generator. The feature library of historical successful levels is input into the discriminator of the generative adversarial network to determine whether the new chessboard layout is reasonable. In response to the reasonableness of the new chessboard layout, the chessboard is regenerated based on the new chessboard layout. Based on the adjusted piece generation strategy, the pieces for the chessboard are generated.

2. The method according to claim 1, characterized in that, The step of extracting features from the operation and the chessboard state respectively to obtain the operation features corresponding to the operation and the chessboard features corresponding to the chessboard state includes: The operation and the chessboard state are divided into units according to a preset time step to obtain multiple operations and chessboard states corresponding to the time step; Feature extraction is performed sequentially on the operation corresponding to each time step and the chessboard state corresponding to each time step to obtain the operation features of the operation corresponding to each time step and the chessboard features of the chessboard state corresponding to each time step.

3. The method according to claim 1, characterized in that, The operational features include operational features corresponding to multiple time steps, and the chessboard features include chessboard features of the chessboard states corresponding to the multiple time steps; determining the object type of the target object based on the operational features and the chessboard features includes: The operation features corresponding to each time step and the chessboard features corresponding to each time step are respectively filtered to obtain the target operation features and target chessboard features corresponding to each time step. According to the order of the time steps, the target operation features and target chessboard features corresponding to each time step are fused sequentially to obtain the first fused feature corresponding to the target operation feature and the second fused feature corresponding to the target chessboard feature. Based on the first fusion feature and the second fusion feature, the object type of the target object is determined.

4. The method according to claim 1, characterized in that, Before determining the agent corresponding to the object type, the method further includes: Based on the operations performed by the agent on the chessboard, the reward weight of the agent is set, and based on the actual operations performed by each agent on the chessboard and the reward weight, the agents are classified to obtain the agents corresponding to each type of object. Obtain the actual success probability of the target object for each of the aforementioned object types; The chessboard is operated by an agent corresponding to each of the object types to determine the simulated success probability of the agent corresponding to each of the object types; Based on the actual success probability and the simulated success probability, an agent corresponding to each object type is trained to obtain a trained agent corresponding to each object type.

5. A processing device for a board game, characterized in that, The device includes: The acquisition module is used to acquire the operations performed by the target object on the chessboard and the state of the chessboard. An extraction module is used to extract features from the operation and the chessboard state respectively, to obtain the operation features corresponding to the operation and the chessboard features corresponding to the chessboard state. The determination module is used to determine the object type of the target object based on the operation characteristics and the chessboard characteristics, and to determine the intelligent agent corresponding to the object type; The execution module is used to simulate the subsequent operations of the target object on the chessboard through the intelligent agent, obtain the success probability of the current piece generation strategy of the chessboard, and count the failure results and step utilization rate of the intelligent agent simulating the execution of the subsequent operations. The step utilization rate is the ratio of the number of steps actually consumed by the intelligent agent in simulating the execution of the subsequent operations to the maximum number of steps of the chessboard. The generation module is used to scan the chessboard according to a preset-sized scanning window based on the failure results, and mark the areas of the chessboard where the failure rate exceeds a preset failure rate threshold as target areas; determine the influence weight of each item combination on the success probability; for each combination, determine a first success probability excluding the item combination and a second success probability including the item combination; determine a first ratio between the second success probability and the first success probability, and subtract the preset value from the first ratio to obtain the conflict coefficient of the item combination on the chessboard; in response to the target area including a piece corresponding to the item combination, increase the conflict coefficient of the item combination by a preset correction value to obtain the corrected conflict coefficient of the item combination; and fuse the influence weight and the corrected conflict coefficient to obtain the influence degree of each item combination on the success probability. In response to the probability of clearing a level being lower than a first threshold, the generation probability of the item combination with a positive influence value is increased to obtain an adjusted piece generation strategy; in response to the probability of clearing a level being higher than a second threshold, the generation probability of the item combination with a negative influence value is increased to obtain an adjusted piece generation strategy; in response to the step utilization rate being lower than a third threshold, the object encoding vector of the target object is obtained, and the heatmap marked with the target area and the object encoding vector of the target object are input into the generator of the generative adversarial network to generate a new chessboard layout corresponding to the chessboard; the feature library of historical successful levels is input into the discriminator of the generative adversarial network to determine whether the new chessboard layout is reasonable, and in response to the new chessboard layout being reasonable, the chessboard is regenerated based on the new chessboard layout; based on the adjusted piece generation strategy, the pieces of the chessboard are generated.

6. An electronic device, characterized in that, The electronic device includes: Memory is used to store executable instructions or computer programs. A processor, configured to execute computer-executable instructions or computer programs stored in the memory, implements the board game processing method according to any one of claims 1 to 4.

7. A computer-readable storage medium storing computer-executable instructions or a computer program, characterized in that, When the computer-executable instructions or computer program are executed by a processor, the method for processing the board game as described in any one of claims 1 to 4 is implemented.

8. A computer program product comprising computer-executable instructions or a computer program, characterized in that, When the computer-executable instructions or computer program are executed by a processor, the method for processing the board game as described in any one of claims 1 to 4 is implemented.

Images