Game processing method and device

Abstract

The invention provides a game processing method and device. The method comprises the steps that in the process that a player plays a game, historical state data of the game and historical operation data triggered by the player for the game are obtained; performing feature extraction on the historical state data to obtain historical state features, and performing feature extraction on the historical operation data to obtain historical operation features; on the basis of the historical state features and the historical operation features, predicting game difficulty when the player continues playing the game to obtain predicted game difficulty; and when the predicted game difficulty is greater than a difficulty threshold, determining prompt information for performing game prompt on the player based on the predicted game difficulty, and displaying the prompt information. According to the method and the device, the game difficulty can be accurately regulated and controlled.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular, to a method and device for processing games. Background Art

[0002] The game program can be any one of a massively multiplayer online role-playing game (MMORPG), a first-person shooting game (FPS), a third-person shooting game, a multiplayer online battle arena game (MOBA), a virtual reality application program, a three-dimensional map program, a card game, a simulation program, or a multiplayer gunfight survival game.

[0003] In the related art, in the current game market, some games provide basic player behavior statistics and recommendation functions through data analysis and simple algorithms. They only recommend game modes based on the player's operation frequency, and cannot predict the challenges that the player may encounter in a specific level or task, and provide targeted prompt information according to these challenges. The recommendation and prompt functions in the related art are relatively basic, unable to accurately predict the game difficulties that the player will encounter, nor can they provide personalized intelligent prompts, resulting in the inability to accurately control the game difficulty. Summary of the Invention

[0004] Embodiments of this application provide a method and device for processing games, an electronic device, a computer-readable storage medium, and a computer program product, which can accurately control the game difficulty.

[0005] The technical solution of the embodiments of this application is implemented as follows:

[0006] Embodiments of this application provide a method for processing games, including:

[0007] During the process of the player playing the game, obtaining the historical state data of the game and the historical operation data triggered by the player for the game;

[0008] Performing feature extraction on the historical state data to obtain historical state features, and performing feature extraction on the historical operation data to obtain historical operation features;

[0009] Based on the historical state features and the historical operation features, predicting the game difficulty when the player continues to play the game to obtain a predicted game difficulty;

[0010] When the predicted game difficulty is greater than the difficulty threshold, based on the predicted game difficulty, determine hint information for giving game hints to the player, and display the hint information.

[0011] An embodiment of the present application provides a processing device for a game, including:

[0012] An acquisition module, configured to acquire historical state data of the game and historical operation data triggered by the player for the game during the process of the player playing the game;

[0013] A feature extraction module, configured to perform feature extraction on the historical state data to obtain historical state features, and perform feature extraction on the historical operation data to obtain historical operation features;

[0014] A prediction module, configured to predict the game difficulty when the player continues to play the game based on the historical state features and the historical operation features, to obtain a predicted game difficulty;

[0015] A display module, configured to, when the predicted game difficulty is greater than the difficulty threshold, determine hint information for giving game hints to the player based on the predicted game difficulty, and display the hint information.

[0016] In the above solution, the acquisition module is further configured to, during the process of the player playing the game, in response to an interaction operation for the game, acquire the interaction duration between the game and the player, and compare the interaction duration with an interaction duration threshold to obtain a duration comparison result; when the duration comparison result indicates that the interaction duration is greater than or equal to the interaction duration threshold, send an acquisition request for the historical state data and the historical operation data to the server; and receive the historical state data and the historical operation data queried by the server for the acquisition request.

[0017] In the above solution, the acquisition module is further configured to, during the process of the player playing the game, respectively perform the following processing at each preset moment: collect historical state data and historical operation data before the preset moment, perform data cleaning on the historical state data and the historical operation data corresponding to the preset moment to obtain cleaned historical state data and historical operation data; and send the cleaned historical state data and historical operation data to the server, so that the server queries the historical state data and the historical operation data when receiving the acquisition request.

[0018] In the above solution, the above-mentioned acquisition module is further configured to, during the process of the player playing the game, perform the following processing respectively at each preset moment: collect historical state data and historical operation data before the preset moment, perform data cleaning on the historical state data and historical operation data corresponding to the preset moment to obtain the cleaned historical state data and historical operation data; and send the cleaned historical state data and historical operation data to the server, so that the server queries the historical state data and the historical operation data when receiving the acquisition request.

[0019] In the above solution, the above-mentioned acquisition module is further configured to, during the process of the player playing the game, in response to an interaction operation for the game, obtain the historical acquisition times of the historical state data and the historical operation data; when the historical acquisition times are zero, obtain the historical state data and the historical operation data; when the historical acquisition times are at least once, obtain the historical acquisition moment of the most recent historical state data and the historical operation data, and determine the time difference between the historical acquisition moment and the current moment; when the time difference is greater than or equal to the time difference threshold, obtain the historical state data and the historical operation data.

[0020] In the above solution, the above-mentioned prediction module is further configured to perform a difficulty prediction on the player based on the historical state feature and the historical operation feature to obtain the prediction probabilities of the player encountering difficulties of each difficulty type; based on the prediction probabilities of the difficulties of each difficulty type, determine the predicted game difficulty when the player continues to play the game.

[0021] In the above solution, the above-mentioned difficulty prediction is implemented through a difficulty prediction model. The difficulty prediction model includes a feature fusion layer and a difficulty prediction layer. The above-mentioned prediction module is further configured to call the feature fusion layer to perform feature fusion on the historical state feature and the historical operation feature to obtain a fusion feature; call the difficulty prediction layer to perform a difficulty prediction on the player based on the fusion feature to obtain the prediction probabilities of the player encountering each of the difficulty types.

[0022] In the above solution, the above-mentioned prediction module is further configured to obtain historical state feature samples and historical operation feature samples, and call an initial difficulty prediction model to perform a difficulty prediction on player samples corresponding to the historical state feature samples and the historical operation feature samples based on the historical state feature samples and the historical operation feature samples to obtain the sample prediction probabilities of the player samples encountering difficulties of each of the difficulty types; for each of the difficulty types, determine the loss value corresponding to the difficulty type based on the sample prediction probability of the difficulty type and the label prediction probability of the difficulty type; based on each of the loss values, train the initial difficulty prediction model to obtain the difficulty prediction model.

[0023] In the above solution, the above prediction module is further configured to compare the predicted probabilities of the difficulties of each of the difficulty types with a probability threshold respectively to obtain the probability comparison results of each of the difficulties; determine the difficulties whose probability comparison results indicate that the predicted probability is greater than the probability threshold as target difficulties; when the number of the target difficulties is one, determine the game difficulty corresponding to the target difficulty as the predicted game difficulty; when the number of the target difficulties is multiple, determine the sum of the game difficulties corresponding to each of the target difficulties as the predicted game difficulty.

[0024] In the above solution, the above display module is further configured to obtain the mapping relationship between a plurality of preset game difficulties and preset prompt information, and compare each of the preset game difficulties in the mapping relationship with the predicted game difficulty respectively to obtain the difficulty comparison results of each of the preset game difficulties; when the difficulty comparison results indicate that the preset game difficulty is the same as the predicted game difficulty, determine the preset prompt information corresponding to the difficulty comparison result in the mapping relationship as the prompt information for giving game prompts to the player.

[0025] An embodiment of the present application provides an electronic device, including:

[0026] A memory for storing computer-executable instructions or computer programs;

[0027] A processor, when executing the computer-executable instructions or computer programs stored in the memory, implements the game processing method provided by the embodiment of the present application.

[0028] An embodiment of the present application provides a computer-readable storage medium storing computer-executable instructions, which when executed by a processor, implement the game processing method provided by the embodiment of the present application.

[0029] An embodiment of the present application provides a computer program product, which includes a computer program or computer-executable instructions, and the computer program or computer-executable instructions are 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 the processor executes the computer-executable instructions, so that the electronic device executes the game processing method described above in the embodiment of the present application.

[0030] The embodiment of the present application has the following beneficial effects:

[0031] During the process of a player playing a game, obtain the historical state data of the game and the historical operation data triggered by the player for the game; extract features from the historical state data to obtain historical state features, and extract features from the historical operation data to obtain historical operation features; based on the historical state features and historical operation features, predict the game difficulty when the player continues to play the game to obtain the predicted game difficulty; when the predicted game difficulty is greater than the difficulty threshold, determine the hint information for giving the player a game hint based on the predicted game difficulty, and display the hint information. In this way, the precise control of the game difficulty is achieved by deeply analyzing the player's behavior and game state. In this process, collect the historical state data and triggered historical operation data of the player in the game, and these data reflect the player's game performance and preferences. By extracting features from these data, the system can obtain the operation features and state features representing the player's game state, and these features are the basis for predicting the player's game difficulty. Then, use these features to predict the difficulty that the player may face when continuing to play the game, so as to obtain the predicted game difficulty. When this predicted difficulty exceeds the preset threshold, corresponding game hint information will be determined and provided according to the prediction result to help the player overcome challenges. It can not only adapt to the user's needs in real time, but also dynamically adjust the game difficulty according to the player's actual performance to achieve precise control. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 is a schematic structural diagram of a system for processing games provided by an embodiment of the present application;

[0033] Figure 2 is a schematic structural diagram of an electronic device for processing games provided by an embodiment of the present application;

[0034] Figure 3 is a schematic flowchart of a method for processing games provided by an embodiment of the present application;

[0035] Figure 4 is a schematic interface diagram of hint information provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be construed as limitations on the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present application.

[0037] In the following description, reference is 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.

[0038] In the following description, the terms "first", "second", and "third" are only used to distinguish similar objects and do not represent a specific order for the objects. It can be understood that, where permitted, "first", "second", and "third" can be interchanged in their specific order or sequence so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0039] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0040] Before further elaborating on the embodiments of this application, the nouns and terms involved in the embodiments of this application are described. The nouns and terms involved in the embodiments of this application are subject to the following explanations.

[0041] 1) Game program: It can be any one of a massively multiplayer online role-playing game (MMORPG), a first-person shooting game (FPS), a third-person shooting game, a multiplayer online battle arena game (MOBA), a virtual reality application, a 3D map program, a simulation program, or a multiplayer gunfight survival game.

[0042] 2) Deep neural network: A deep neural network (DNN for short) is a complex machine learning model that mimics the way neurons in the human brain are connected and learns and extracts deep features of data through multiple layers of non-linear processing units. A deep neural network is a neural network with a multi-layer structure, where each layer contains multiple neurons that are connected by weighted connections and process data through non-linear activation functions. A deep neural network is a powerful machine learning model that can achieve high-performance prediction and classification on various data through its multi-layer structure and complex non-linear processing capabilities.

[0043] 3) Convolutional Neural Networks (CNN): It is a type of feed-forward neural network (FNN) that contains convolutional calculations and has a deep structure, and is one of the representative algorithms of deep learning. Convolutional neural networks have the ability of representation learning and can perform shift-invariant classification on input images according to their hierarchical structure.

[0044] 4) Artificial Intelligence (AI): It is an interdisciplinary subject that covers a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, pre-trained model technology, operation / interaction systems, mechatronics, etc. Among them, the pre-trained model is also known as the large model or the basic model, and can be widely applied to downstream tasks in various directions of artificial intelligence after fine-tuning. The software technologies of artificial intelligence mainly include several major directions such as computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0045] During the implementation process of the embodiments of this application, the applicant found the following problems in the related technologies:

[0046] In the related technologies, in the current game market, some games have provided basic player behavior statistics and recommendation functions through data analysis and simple algorithms. They only recommend game modes based on the player's operation frequency, but cannot predict the challenges that players may encounter in specific levels or tasks and provide targeted prompt information according to these challenges. The recommendation and prompt functions in the related technologies are relatively basic, unable to accurately predict the game difficulties that players will encounter, nor can they provide personalized intelligent prompts, resulting in the inability to achieve accurate regulation of game difficulty.

[0047] The embodiments of this application provide a game processing method, device, electronic device, computer-readable storage medium, and computer program product, which can achieve accurate regulation of game difficulty. The following describes the exemplary application of the game processing system provided by the embodiments of this application.

[0048] See Figure 1 , Figure 1 It is a schematic diagram of the architecture of the game processing system 100 provided by the embodiments of this application. The terminal (exemplarily shows the terminal 400) is connected to the server 200 through the network 300. The network 300 can be a wide area network, a local area network, or a combination of both.

[0049] The terminal 400 is used for a user to use the client 410 to display a prompt message on the graphical interface 410-1 (the graphical interface 410-1 is exemplarily shown). The terminal 400 and the server 200 are interconnected through a wired or wireless network.

[0050] In some embodiments, the server 200 may be an independent physical server, or a server cluster or business system composed of multiple physical servers. It may also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms. The terminal 400 may be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart TV, a smart watch, a vehicle-mounted terminal, etc., but is not limited thereto. The electronic device provided in the embodiments of the present application may be implemented as a terminal or as a server. The terminal and the server may be directly or indirectly connected through a wired or wireless communication method, which is not limited in the embodiments of the present application.

[0051] In some embodiments, during the process of the player playing the game, the server 200 obtains the historical state data of the game and the historical operation data triggered by the player for the game; extracts features from the historical state data to obtain historical state features, and extracts features from the historical operation data to obtain historical operation features; based on the historical state features and the historical operation features, predicts the game difficulty when the player continues to play the game to obtain a predicted game difficulty, and sends the predicted game difficulty to the terminal 400. When the predicted game difficulty is greater than a difficulty threshold, the terminal 400 determines a prompt message for prompting the player based on the predicted game difficulty and displays the prompt message.

[0052] In some other embodiments, during the process of the player playing the game, the server 200 obtains the historical state data of the game and the historical operation data triggered by the player for the game; extracts features from the historical state data to obtain historical state features, and extracts features from the historical operation data to obtain historical operation features; based on the historical state features and the historical operation features, predicts the game difficulty when the player continues to play the game to obtain a predicted game difficulty. When the predicted game difficulty is greater than a difficulty threshold, a prompt message for prompting the player is determined based on the predicted game difficulty, and the prompt message is sent to the terminal 400, and the terminal 400 displays the prompt message.

[0053] See also Figure 2 , Figure 2 is a schematic diagram of the structure of an electronic device 500 for processing games provided in an embodiment of the present application, wherein: Figure 2 The electronic device 500 shown may be Figure 1 The server 200 or the terminal 400 in Figure 2 The electronic device 500 shown includes: at least one processor 430, a memory 450, and at least one network interface 420. The various components in the electronic device 500 are coupled together via a bus system 440. It is understood that the bus system 440 is used to achieve connection and communication between these components. In addition to the data bus, the bus system 440 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, the bus system 440 is not described in detail. Figure 2 Various buses are labeled as bus system 440 .

[0054] The processor 430 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., where the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0055] The memory 450 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state memory, hard drives, optical drives, etc. The memory 450 may optionally include one or more storage devices that are physically remote from the processor 430.

[0056] The memory 450 includes a volatile memory or a non-volatile memory, and may also include both volatile and non-volatile memories. The non-volatile memory may be a read-only memory (ROM), and the volatile memory may be a random access memory (RAM). The memory 450 described in the embodiments of the present application is intended to include any suitable type of memory.

[0057] In some embodiments, memory 450 can store data to support various operations, examples of which include programs, modules, and data structures, or a subset or superset thereof, as exemplarily described below.

[0058] Operating system 451, including system programs for processing various basic system services and performing hardware-related tasks, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks;

[0059] A network communication module 452 is used to reach other electronic devices via one or more (wired or wireless) network interfaces 420. Exemplary network interfaces 420 include: Bluetooth, Wireless Fidelity (WiFi), Universal Serial Bus (USB), etc.

[0060] In some embodiments, the processing device of the game provided by the embodiments of the present application can be implemented in software. Figure 2 The processing device 455 of the game stored in the memory 450 is shown. It can be software in the form of programs and plugins, etc., and includes the following software modules: an acquisition module 4551, a feature extraction module 4552, a prediction module 4553, and a display module 4554. These modules are logical, so they can be combined arbitrarily or further split according to the functions implemented. The functions of each module will be described below.

[0061] In other embodiments, the processing device of the game provided by the embodiments of the present application can be implemented in hardware. As an example, the processing device of the game provided by the embodiments of the present application can be a processor in the form of a hardware decoding processor, which is programmed to execute the game processing method provided by the embodiments of the present application. For example, a processor in the form of a hardware decoding processor can employ one or more Application Specific Integrated Circuits (ASICs), DSPs, Programmable Logic Devices (PLDs), Complex Programmable Logic Devices (CPLDs), Field-Programmable Gate Arrays (FPGAs), or other electronic components.

[0062] In some embodiments, the terminal or server can implement the game processing method provided by the embodiments of the present application by running a computer program or computer-executable instructions. For example, the computer program can be a native program in the operating system (e.g., a dedicated game program) or a software module. For example, it can be a game module embedded in any program (such as an instant messaging client, a photo album program, an electronic map client, a navigation client); for example, it can be a Native Application (APP), that is, a program that needs to be installed in the operating system to run. In short, the above computer program can be any form of application program, module, or plugin.

[0063] The exemplary applications and implementations of the server or terminal provided in the embodiments of the present application will be combined to illustrate the game processing method provided in the embodiments of the present application.

[0064] See Figure 3 , Figure 3 which is a schematic flowchart of the game processing method provided in the embodiments of the present application. It will be described in combination with Figure 3 the steps 101 to 104 shown. The game processing method provided in the embodiments of the present application can be implemented independently by the server or the terminal, or jointly implemented by the server and the terminal. Below, it will be described by taking the server implementing independently as an example.

[0065] In step 101, during the process of the player playing the game, obtain the historical status data of the game and the historical operation data triggered by the player for the game.

[0066] In some embodiments, the historical status data refers to dynamic information such as the game environment, level status, resource distribution, and enemy behavior recorded during the operation of the game. These data reflect the overall state of the game at a certain point in time. Including but not limited to the progress of the game level, the player's current health value, the quantity of resources, the position of enemies, the completion status of tasks, etc. By analyzing the historical status data, the change trend of the game environment can be understood, providing a basis for predicting the future game difficulty of the player.

[0067] In some embodiments, the historical operation data refers to all operation records triggered by the player during the game process, including key input, mouse click, skill release, movement path, etc. These data reflect the player's behavior habits and operation skills. Including the player's movement trajectory, attack frequency, skill usage order, item usage situation, etc. By analyzing the historical operation data, the player's operation level, strategy selection, and understanding of the game mechanism can be evaluated, thereby providing a basis for personalized difficulty adjustment.

[0068] In some embodiments, the above step 101 can be implemented in the following manner: during the process of the player playing the game, in response to the interaction operation for the game, obtain the interaction duration between the game and the player, and compare the interaction duration with the interaction duration threshold to obtain a duration comparison result; when the duration comparison result indicates that the interaction duration is greater than or equal to the interaction duration threshold, send a request for obtaining the historical status data and the historical operation data to the server; receive the historical status data and the historical operation data queried by the server for the obtaining request.

[0069] In some embodiments, during the process of a player playing a game, the system dynamically determines whether it is necessary to obtain more detailed historical data (including historical state data and historical operation data) by monitoring the interaction duration between the player and the game and combining it with a preset interaction duration threshold. The core purpose of this mechanism is to trigger data acquisition and analysis at an appropriate time, so as to provide support for subsequent difficulty prediction, hint information generation, or personalized game experience optimization. The system monitors the player's interaction operations with the game in real time (such as key presses, clicks, movements, etc.) and records the interaction duration. The interaction duration reflects the player's level of engagement and concentration in the current game session. The recorded interaction duration is compared with the preset interaction duration threshold. The interaction duration threshold is a preset time value used to determine whether the player has fully participated in the current game session. If the interaction duration is greater than or equal to the threshold, it indicates that the player has had a relatively long game interaction, and at this time, it is considered necessary to obtain more detailed historical data to support further analysis. A request is sent to the server to obtain historical state data and historical operation data. If the interaction duration is greater than or equal to the threshold, it indicates that the player has had a relatively long game interaction, and at this time, it is considered necessary to obtain more detailed historical data to support further analysis. The system sends a request to the server to obtain historical state data and historical operation data. The server queries and returns the corresponding historical data according to the request. The historical state data includes information such as the game environment and level status, and the historical operation data includes the player's operation records and behavior characteristics.

[0070] In some embodiments, the interaction duration is an important indicator for judging the player's participation. A longer interaction duration usually means that the player has a higher level of engagement in the current game session, and at this time, obtaining historical data is more meaningful. By setting a reasonable interaction duration threshold, frequent data requests can be avoided, reducing the system burden. The data acquisition request is only triggered when the interaction duration reaches the threshold, which is a mechanism for obtaining data on demand and can improve the efficiency and response speed of the system. The server is responsible for storing and querying historical data to ensure the integrity and timeliness of the data. Through the data returned by the server, the system can conduct more in-depth analysis, such as difficulty prediction, behavior pattern recognition, etc. When the player stays at a certain level for a long time, the system can adjust the difficulty of subsequent levels based on historical data to match the player's skill level. If the player encounters difficulties in a certain session, the system can generate targeted hint information based on the historical operation data to help the player clear the level smoothly. By analyzing the historical state data and operation data, developers can understand the player's behavior patterns and optimize the game design.

[0071] As an example, assume that the player is playing an online card game. The system dynamically adjusts the game experience or provides prompt information by monitoring the player's interaction duration and historical data. The player continuously performs interaction operations in a certain game, such as moving cards, choosing to play cards, clicking the prompt button, etc. The system records the interaction duration of the player from the start of the game to the current moment. Assume that the preset interaction duration threshold of the system is 5 minutes. If the player has continuously interacted for 5 minutes or longer in the current game, the system will trigger the next operation. When the interaction duration reaches or exceeds 5 minutes, the system sends a request to the server to obtain the player's historical status data and historical operation data. The card face distribution of the current game round (such as the remaining card deck, the played cards, the player's hand cards, etc.). The game progress (such as the current level, the remaining time, etc.). The historical operation data may include: the player's card-playing order, the frequency of using the prompt function, the number of times of moving cards, etc. The server queries and returns the relevant data according to the request. After receiving the data, the system can further analyze the player's behavior pattern and the current game state.

[0072] In this way, during the player's game process, by detecting the player's interaction duration with the game and comparing it with the preset interaction duration threshold, the acquisition request for the historical status data and historical operation data can be automatically triggered after the player has invested sufficient time. This mechanism can effectively identify the player's engagement and potential difficulties in the game, thus providing data support for subsequent analysis and optimization. For example, when the player stays at a certain level for too long, prompt information can be generated or the game difficulty can be adjusted based on the historical data to help the player pass the level smoothly. This way of obtaining data on demand not only improves the efficiency of the system, but also provides a more personalized and smooth game experience for the player, and at the same time provides a scientific basis for developers to optimize the game design, ultimately enhancing the user satisfaction and stickiness of the game.

[0073] In some embodiments, before obtaining the historical status data of the game and the historical operation data triggered by the player for the game, the following processing may also be performed: During the player's game process, the following processing is respectively performed at each preset moment: Collect the historical status data and historical operation data before the preset moment, perform data cleaning on the historical status data and historical operation data corresponding to the preset moment to obtain the cleaned historical status data and historical operation data; and send the cleaned historical status data and historical operation data to the server so that the server queries the historical status data and the historical operation data when receiving the acquisition request.

[0074] In some embodiments, before obtaining historical status data and historical operation data, the system adds steps of data collection, cleaning, and pre-storage. The purpose of this preprocessing process is to ensure the quality and availability of data, while improving the efficiency of subsequent queries and analysis. By regularly collecting, cleaning, and uploading data at preset times, the system can provide more accurate and complete historical data to the server, so as to quickly respond to players' requests when needed. During the game process, historical status data and historical operation data are collected regularly (such as every 5 minutes or after completing each level). Historical status data includes information such as the game environment, level status, resource distribution, etc. Historical operation data includes information such as players' operation records and behavior habits. The collected data is cleaned to remove invalid, duplicate, or incorrect data. Specific operations for data cleaning may include: removing outliers (such as values outside a reasonable range). Filling in missing values (such as using the average value or interpolation method to fill in missing operation records). Standardizing data formats (such as unifying the timestamp format or operation type encoding). The cleaned data is more accurate and consistent, and is suitable for subsequent analysis and queries. The cleaned historical status data and historical operation data are uploaded to the server. The server stores these data in the database for subsequent queries. When the system needs to obtain historical data (such as when the interaction duration reaches a threshold), the server can quickly query and return the cleaned data. Since the data has been cleaned and pre-stored, the query efficiency is higher and the response speed is faster. By regularly collecting, cleaning, and uploading historical status data and historical operation data at preset times, the system can ensure the quality and availability of data, while improving the efficiency of subsequent queries and analysis. This preprocessing process provides reliable data support for the intelligent functions of the game (such as difficulty adjustment, hint generation, etc.), and ultimately improves the players' gaming experience and the overall performance of the system.

[0075] In some embodiments, the preset time refers to a time point or event trigger point preset during the game process for regularly or conditionally collecting data. These time points can be fixed time intervals (such as every 5 minutes) or specific game events (such as completing a level, triggering a certain task, etc.). Through the preset time, the system can collect historical status data and historical operation data in a planned manner, ensuring the continuity and integrity of the data, while avoiding the performance overhead that may be brought by real-time collection.

[0076] In some embodiments, data cleaning refers to processing the collected raw data to remove invalid, duplicate, incorrect, or inconsistent data, thereby improving the quality and availability of the data. Data cleaning usually includes operations such as removing outliers, filling in missing values, and standardizing data formats. Data cleaning can ensure the accuracy of subsequent analysis and queries, avoid the influence of noise data on the results, and at the same time improve the storage and query efficiency of the data.

[0077] In some embodiments, assume that a player is playing an online solitaire game (such as Spider Solitaire). The system regularly collects, cleans, and uploads historical state data and historical operation data at preset times to quickly respond to the player's requests when needed. The system sets the preset time to every completed game or every 5 minutes (whichever comes first). For example: The player completes a game of Spider Solitaire. The player has interacted for more than 5 minutes in a certain game. At the preset time, the system collects the following data: Historical state data: The card face distribution of the current game (such as the remaining deck, the played cards, the player's hand cards, etc.). The game progress (such as the current level, the remaining time, etc.). Historical operation data: The player's card-playing order, the frequency of using the hint function, the number of times of moving cards, etc. Clean the collected data to ensure the accuracy and consistency of the data: Remove outliers: For example, delete the abnormal operation records where the player clicks hundreds of times continuously in a very short time. Fill in missing values: For example, if the timestamps of some operation records are missing, use interpolation to fill them. Standardize the data: For example, unify the timestamp format to "YYYY-MM-DD HH:MM:SS," and encode the operation types into a unified format (such as "Move cards" is encoded as "MOVE"). Upload the cleaned historical state data and historical operation data to the server. The server stores this data in the database for subsequent queries.

[0078] In this way, by regularly collecting, cleaning, and uploading historical state data and historical operation data at preset times during the player's game process, the system can ensure the accuracy, integrity, and availability of the data. Data cleaning removes invalid and incorrect information, improving the data quality, while the pre-storage mechanism significantly improves the efficiency and response speed of server queries. This preprocessing process provides reliable data support for subsequent difficulty prediction, hint generation, and personalized game experience optimization, while reducing the burden of real-time data processing. Ultimately, this mechanism can not only enhance the player's game experience but also provide a scientific basis for developers to optimize game design, enhance user stickiness and satisfaction.

[0079] In some other embodiments, during the process of the player playing the game, obtaining the historical state data of the game and the historical operation data triggered by the player for the game can also be achieved in the following manner: during the process of the player playing the game, in response to an interaction operation for the game, obtain the historical acquisition times of the historical state data and the historical operation data; when the historical acquisition times are zero, obtain the historical state data and the historical operation data; when the historical acquisition times are at least once, obtain the historical acquisition time of the most recent historical state data and the historical operation data, and determine the time difference between the historical acquisition time and the current time; when the time difference is greater than or equal to the time difference threshold, obtain the historical state data and the historical operation data.

[0080] In some embodiments, by introducing the concepts of historical acquisition times and time difference threshold, the acquisition logic of historical state data and historical operation data is optimized. This design can avoid repeated data acquisition, reduce waste of system resources, and ensure the timeliness of data. By dynamically judging whether to re-acquire data, the system can improve operation efficiency on the premise of ensuring data accuracy. Record the acquisition times of historical state data and historical operation data during the game process of the player. If the historical acquisition times are zero, it means that the relevant data has not been acquired yet, and at this time, it is necessary to acquire immediately. Directly acquire the historical state data and the historical operation data. The historical acquisition times are at least once: the system obtains the time of the most recent data acquisition and calculates the time difference between this time and the current time. If the time difference is greater than or equal to the preset time difference threshold, it means that the data may have become outdated and needs to be re-acquired. If the time difference is less than the threshold, it means that the data is still valid and there is no need to re-acquire. When the conditions are met (the historical acquisition times are zero or the time difference exceeds the threshold), the system acquires the latest historical state data and historical operation data. By recording the historical acquisition times, the system can judge whether it is necessary to acquire data for the first time and avoid repeated operations. For the case where data has been acquired, the system can further judge the timeliness of the data. The time difference threshold is used to judge whether the data needs to be updated. If the data acquisition time is too far from the current time, it may have become invalid and needs to be re-acquired. Reasonably setting the time difference threshold can balance the timeliness of data and system performance.

[0081] As an example, assume that the player is playing Spider Solitaire, and the system records the historical acquisition count and the time difference threshold (e.g., 10 minutes): If the player has just started the game and the historical acquisition count is zero, the system immediately acquires the historical status data and historical operation data. If the player has been playing for some time and the last data acquisition was 8 minutes ago (the time difference is less than the threshold), the system considers the data still valid and does not need to re-acquire it. If the last data acquisition was 15 minutes ago (the time difference exceeds the threshold), the system re-acquires the latest historical status data and historical operation data to ensure the timeliness of the data.

[0082] In this way, by dynamically judging the acquisition timing of the historical status data and historical operation data during the player's game process, the system can effectively avoid repeated data acquisition, reduce resource waste, and at the same time ensure the timeliness of the data. When the historical acquisition count is zero, the system immediately acquires data to meet the initial requirements; when the historical acquisition count is at least one, the system determines whether to re-acquire data by comparing the time difference between the last acquisition time and the current time. This mechanism not only improves the operating efficiency of the system but also provides accurate and timely data support for subsequent intelligent functions such as difficulty adjustment and hint generation, thereby enhancing the player's gaming experience and the overall performance of the system.

[0083] In step 102, feature extraction is performed on the historical status data to obtain historical status features.

[0084] In some embodiments, the historical status data reflects the overall state of the game environment at a certain point in time, such as the level progress, resource distribution, enemy behavior, etc. Key features that can characterize the changes in the game environment are extracted from the historical status data. The completion percentage of the current level, the remaining number of tasks, etc. The remaining health points, item quantity, gold quantity, etc. of the player. The number, location, attack frequency, etc. of the enemies. The length of time the game has been played, the time consumed in the current level, etc. The feature extraction process can be the extraction of statistical features: calculating the mean, variance, maximum value, minimum value, etc. It can also be the extraction of trend features: analyzing the change trend of the data, such as the resource consumption speed, the level progress change rate, etc. It can also be the extraction of classification features: dividing the data into different categories, such as the level difficulty level, the resource scarcity degree, etc.

[0085] In step 103, feature extraction is performed on the historical operation data to obtain historical operation features.

[0086] In some embodiments, the historical operation data records the specific operations of the player in the game, such as button inputs, skill releases, movement paths, etc. Key features that can characterize the player's behavior habits and skill levels are extracted from the historical operation data. It can also be the extraction of statistical features: calculating the number of operations, success rate, error rate, etc. It can also be the extraction of sequence features: analyzing the patterns of operation sequences, such as using Markov chain models or sequence matching algorithms. It can also be the extraction of clustering features: classifying the player's operation behaviors into different categories, such as aggressive type, conservative type, etc. By extracting key features, the original data is transformed into a more concise and representative form, reducing the complexity of data processing. The data after feature extraction is more suitable for training and prediction of machine learning models, and can significantly improve the analysis efficiency. High-quality features can improve the accuracy and generalization ability of the model, providing reliable support for subsequent intelligent functions.

[0087] In step 104, based on the historical state features and the historical operation features, the game difficulty when the player continues to play the game is predicted to obtain the predicted game difficulty.

[0088] In some embodiments, by analyzing the player's behavior patterns and changes in the game environment, the system can dynamically evaluate the player's performance in subsequent games and predict the challenges they may face. This prediction mechanism can provide a scientific basis for game difficulty adjustment, hint generation, etc., thereby enhancing the player's gaming experience. Historical state features: Key features reflecting the game environment, such as level progress, resource distribution, enemy behavior, etc. Historical operation features: Key features reflecting the player's behavior habits and skill levels, such as operation frequency, operation sequence, operation accuracy, etc. The historical state features and historical operation features are integrated into a comprehensive feature vector as the input of the prediction model. A machine learning model (such as a regression model, classification model, or deep learning model) is used to analyze the integrated features to predict the player's performance and possible difficulties in subsequent games. The output of the model is the predicted game difficulty, usually represented in the form of a numerical value or a level (such as difficulty level 1-10). According to the predicted game difficulty, the system can dynamically adjust the difficulty of subsequent levels or generate targeted hint information to help the player better cope with challenges. The predicted game difficulty is a dynamic assessment based on current and historical data and can be adjusted in real time as the player's behavior changes. This dynamic nature ensures the accuracy and adaptability of the difficulty prediction.

[0089] In some embodiments, if the predicted game difficulty is high, the difficulty of subsequent levels can be reduced, such as reducing the number of enemies or increasing the resource supply. If the predicted game difficulty is low, the system can increase the difficulty of subsequent levels to maintain the challenge of the game. If the predicted game difficulty is high, the system can generate targeted hint information to help the player improve their operation or adjust their strategy. For example, in a card game, the system can prompt the player to prioritize clearing a certain pile of cards. By analyzing the prediction results, developers can understand the player's behavior patterns and preferences, thereby optimizing the game design. For example, adjusting the level design, resource distribution, or enemy behavior to better match the player's skill level. By predicting the game difficulty based on historical state features and historical operation features, and integrating the key features of the game environment and player behavior, the system can accurately evaluate the player's performance in subsequent games and dynamically adjust the difficulty or generate hint information. This mechanism can not only enhance the player's gaming experience but also provide a basis for developers to optimize the game design, ultimately enhancing the attractiveness and user stickiness of the game.

[0090] In some embodiments, step 104 above can be implemented in the following manner: Based on the historical state features and the historical operation features, perform a difficulty prediction on the player to obtain the predicted probabilities of the player encountering difficulties of each difficulty type; Based on the predicted probabilities of the difficulties of each difficulty type, determine the predicted game difficulty when the player continues to play the game.

[0091] In some embodiments, by predicting the probabilities of the player encountering difficulties under different difficulty types, the system can more accurately evaluate the player's ability level and the possible challenges they may face, thereby dynamically adjusting the game difficulty. This mechanism can not only enhance the personalized experience of the game but also help the player better adapt to the game rhythm. Use classification models (such as logistic regression, random forest, neural network, etc.) to analyze the integrated features and predict the probabilities of the player encountering difficulties under different difficulty types. The output of the model is the predicted probabilities of the difficulties of each difficulty type. For example: The probability of encountering difficulties at low difficulty: 10%. The probability of encountering difficulties at medium difficulty: 50%. The probability of encountering difficulties at high difficulty: 90%. Based on the predicted probabilities of the difficulties of each difficulty type, the system can comprehensively evaluate the player's ability level and determine the predicted game difficulty suitable for the player. For example, if the probability of the player encountering difficulties at high difficulty is high (such as 90%), the difficulty of subsequent levels can be reduced. By predicting the probabilities of the player encountering difficulties under different difficulty types, the system can more accurately evaluate the player's ability level and the possible challenges they may face. The fine-grained prediction provides a more scientific basis for dynamic difficulty adjustment. Commonly used classification models include logistic regression, random forest, support vector machine (SVM), etc. For complex scenarios, deep learning models (such as neural networks) can be used to improve the prediction accuracy.

[0092] In some embodiments, the difficulty of subsequent levels can be dynamically adjusted according to the difficulty prediction probability and a preset difficulty threshold. For example, if the probability that a player encounters difficulties at a high difficulty level exceeds a certain threshold (such as 80%), the system can lower the difficulty. If the probability that a player encounters difficulties at a high difficulty level is relatively high, the difficulty of subsequent levels can be reduced, such as reducing the number of enemies or increasing the resource supply. If the probability that a player encounters difficulties at a low difficulty level is relatively low, the system can increase the difficulty of subsequent levels to maintain the challenge of the game. If the probability that a player encounters difficulties in a certain difficulty type is relatively high, the system can generate targeted hint information to help the player improve their operations or adjust their strategies. For example, in a card game, the system can prompt the player to prioritize clearing a certain card pile. By analyzing the difficulty prediction probability, developers can understand the player's behavior patterns and preferences, and thus optimize the game design. For example, adjusting the level design, resource distribution, or enemy behavior to better match the player's skill level.

[0093] As an example, assume that the player is playing Spider Solitaire, and the system predicts the probability that the player encounters difficulties in different difficulty types based on historical state features and historical operation features: The card distribution of the current game: There are relatively many remaining card piles and few movable cards. Game progress: 70% of the current level has been completed, and the remaining time is short. Operation frequency: The number of times the player moves cards per minute is relatively low. Operation accuracy: The player frequently tries incorrect card combinations. The probability of encountering difficulties at a low difficulty level: 20%. The probability of encountering difficulties at a medium difficulty level: 60%. The probability of encountering difficulties at a high difficulty level: 90%. Since the probability that the player encounters difficulties at a high difficulty level is relatively high (90%), the system adjusts the difficulty of the subsequent level to medium difficulty.

[0094] In this way, it is possible to more accurately predict the player's game performance and possible difficulties, thereby providing a scientific basis for dynamically adjusting the game difficulty and generating personalized hint information. This prediction method based on data and models can not only enhance the player's game experience, but also help developers understand the player's behavior patterns, further optimize the game design, and enhance the attractiveness and user stickiness of the game. Based on the probability distribution of difficulty prediction, the system can implement a more flexible and refined game difficulty adjustment strategy to ensure that the game can challenge the player's abilities while avoiding overly frustrating the player's enthusiasm, thereby improving the player's satisfaction and retention rate while maintaining the fun of the game. Through this mechanism, the game can better adapt to the player's needs, achieve a positive interaction between the player and the game, and promote the continuous development and innovation of the game industry.

[0095] In some embodiments, the difficulty prediction is implemented through a difficulty prediction model, and the difficulty prediction model includes a feature fusion layer and a difficulty prediction layer.

[0096] In some embodiments, the above-mentioned prediction of difficulties for the player based on the historical state features and the historical operation features to obtain the prediction probabilities of the player encountering difficulties of various difficulty types can be achieved in the following manner: Call the feature fusion layer to fuse the historical state features and the historical operation features to obtain fused features; call the difficulty prediction layer to predict the difficulties for the player based on the fused features to obtain the prediction probabilities of the player encountering each of the difficulty types.

[0097] In some embodiments, the main task of the feature fusion layer is to integrate the historical state features and the historical operation features into a unified fused feature. Concatenate the historical state features and the historical operation features in the feature dimension to form a high-dimensional feature vector. Assign different weights to the historical state features and the historical operation features according to the importance of the features, and then perform weighted summation. Use a dense layer or an attention mechanism to perform non-linear fusion on the features. The fused feature is a high-dimensional vector that can simultaneously reflect the changes in the game environment and the player's behavior pattern.

[0098] In some embodiments, the difficulty prediction layer predicts the probabilities of the player encountering difficulties under different difficulty types based on the fused features. The difficulty prediction layer is usually a classification model, such as a dense neural network or a softmax classifier. The output of the difficulty prediction layer is a probability distribution representing the probabilities of the player encountering difficulties under different difficulty types. For example: The probability of encountering difficulties at low difficulty: 10%. The probability of encountering difficulties at medium difficulty: 50%. The probability of encountering difficulties at high difficulty: 90%.

[0099] As an example, the input data: Historical state features: Key features reflecting the game environment. Historical operation features: Key features reflecting the player's behavior habits and skill levels. Call the feature fusion layer to integrate the historical state features and the historical operation features into fused features. Call the difficulty prediction layer to predict the probabilities of the player encountering difficulties under different difficulty types based on the fused features. Output the difficulty prediction probabilities for each difficulty type, such as the probability distributions at low, medium, and high difficulties.

[0100] In some embodiments, the feature fusion layer can integrate different types of data (historical state features and historical operation features) into a unified representation, thereby improving the expressive power of the model. By fusing features, the model can simultaneously consider the changes in the game environment and the player's behavior pattern, providing more comprehensive information for difficulty prediction. The output of the difficulty prediction layer is a probability distribution, which can reflect the player's performance under different difficulty types. The fine-grained prediction provides a scientific basis for dynamic difficulty adjustment and personalized prompt generation. The designs of the feature fusion layer and the difficulty prediction layer have high scalability and can adjust the model structure according to specific requirements (such as adding an attention mechanism or using a more complex neural network).

[0101] Thus, the difficulty prediction model adopting the feature fusion layer and the difficulty prediction layer can significantly improve the accuracy and efficiency of prediction by fusing the historical state features and the historical operation features and then performing difficulty prediction. It can not only better capture the complex relationship between the player's behavior and the game environment, but also provide strong support for the game's dynamic difficulty adjustment and personalized experience. The feature fusion layer effectively integrates data from different sources, enabling the model to comprehensively understand the player's game performance from multiple dimensions, while the difficulty prediction layer, based on these fused features, accurately predicts the difficulties that the player may encounter, so that the game can respond to the player's needs in a timely manner and provide a more considerate game experience. This helps to improve the player's game satisfaction, increase user stickiness, and at the same time provides a data basis for game developers to deeply understand the player's behavior, provides a direction for game optimization and iteration, and promotes the innovative development of the game industry.

[0102] In some embodiments, before predicting the difficulties that the player encounters for each difficulty type based on the historical state features and the historical operation features to obtain the prediction probabilities, the following processing can also be performed: Obtain historical state feature samples and historical operation feature samples, and call the initial difficulty prediction model. Based on the historical state feature samples and the historical operation feature samples, perform difficulty prediction on the player samples corresponding to the historical state feature samples and the historical operation feature samples to obtain the sample prediction probabilities of the player samples encountering the difficulties of each difficulty type; for each difficulty type, determine the loss value corresponding to the difficulty type based on the sample prediction probability of the difficulty type and the label prediction probability of the difficulty type; based on each loss value, train the initial difficulty prediction model to obtain the difficulty prediction model.

[0103] In some embodiments, before implementing the difficulty prediction model to obtain the predicted probabilities of players encountering difficulties of various difficulty types, the following steps are necessary preprocessing and model training processes: Collect a series of historical state feature samples in the game, which reflect players' performance and environmental changes in different game states. At the same time, collect historical feature samples of players, which record players' behaviors and operation habits. Based on the historical state features and historical operation features, construct a preliminary difficulty prediction model. This model may be a simple statistical model or a deep learning model. Set initial values for the parameters of the model, which will be adjusted in the subsequent process. Input the obtained historical state feature samples and historical operation feature samples into the initial difficulty prediction model. The model will output the predicted probabilities of players' samples encountering difficulties under various difficulty types. Corresponding to the sample predicted probabilities, there is a label predicted probability, which represents the probability of players actually encountering difficulties. For each difficulty type, calculate the difference between the sample predicted probability and the label predicted probability, that is, the loss value. This is usually achieved through a loss function (such as cross-entropy loss). Use the calculated loss value for the backpropagation process of the model to reduce the loss value by adjusting the model parameters. Repeat the above process multiple times, adjusting the model parameters in each iteration until the loss value of the model reaches a satisfactory level or no longer decreases significantly. Use an independent validation set to test the trained model to ensure that the model has good generalization ability. Evaluate the performance of the model, such as accuracy, recall rate, F1 score, etc. Deploy the trained difficulty prediction model to the game system to predict the difficulty probability in real time during players' gaming process. As the game progresses, continue to collect new historical state feature samples and historical operation feature samples. Regularly update the model with new data to maintain the accuracy and adaptability of the model.

[0104] Thus, by obtaining historical state feature samples and historical operation feature samples and using these samples to train the initial difficulty prediction model, the prediction accuracy and generalization ability of the model can be significantly improved. This method enables the model to better understand and learn the complex relationship between players' behavior patterns and the game environment, thus being more accurate in predicting the likelihood of players encountering difficulties. In addition, by calculating the loss value between the sample predicted probability and the label predicted probability and training the model specifically, the model parameters can be effectively optimized and the prediction error can be reduced.

[0105] In some embodiments, to determine the predicted game difficulty when the player continues to play the game based on the predicted probabilities of difficulties of each of the difficulty types, the following processing may also be performed: comparing the predicted probabilities of difficulties of each of the difficulty types with a probability threshold respectively to obtain the probability comparison results of each of the difficulties; determining the difficulties whose probability comparison results indicate that the predicted probabilities are greater than the probability threshold as target difficulties; when the number of the target difficulties is one, determining the game difficulty corresponding to the target difficulty as the predicted game difficulty; and when the number of the target difficulties is multiple, determining the sum of the game difficulties corresponding to each of the target difficulties as the predicted game difficulty.

[0106] In some embodiments, the predicted probabilities of difficulties of each difficulty type are compared with a preset probability threshold. This threshold represents the critical point at which a player encounters difficulties. Difficulty types with predicted probabilities exceeding this threshold are considered difficulties that the player may encounter. The comparison results will indicate which difficulty types have predicted probabilities exceeding the probability threshold, and these difficulty types are considered the target difficulties that the player may encounter. Based on the probability comparison results, identify the difficulties whose predicted probabilities are greater than the probability threshold, and these are called target difficulties. If there is only one target difficulty, then the game difficulty corresponding to this difficulty will be determined as the predicted game difficulty. If there are multiple target difficulties, then the game difficulties corresponding to these target difficulties are summed up, and the resulting sum will be determined as the predicted game difficulty. The setting of the probability threshold is crucial as it determines which difficulty types will be considered target difficulties. The threshold needs to be determined according to the specific situation of the game and the average ability level of the players. When there is only one target difficulty, the predicted game difficulty directly corresponds to the game difficulty of that difficulty. When there are multiple target difficulties, the predicted game difficulty is determined by summing up, which assumes that the impacts of all target difficulties on the game difficulty are equivalent. In the case of multiple target difficulties, the summing method provides a simple way to determine the predicted game difficulty. Further analysis is needed to ensure its rationality because different target difficulties may have different degrees of impact on the game difficulty.

[0107] As an example, in the application scenario of a card game, the following is an illustration of how to use the described processing method to determine the predicted game difficulty when a player continues to play the game: Suppose a player is playing a card game, such as Spider Solitaire. The difficulty types in the game may include "Easy", "Medium", and "Hard". The system has predicted the probabilities of the player encountering difficulties under each difficulty type based on the player's historical state characteristics and historical operation characteristics. The system sets a probability threshold, for example, 0.5, which means that if the predicted probability of a difficulty type exceeds 50%, then this difficulty type will be considered the target difficulty that the player may encounter. The system compares the predicted probabilities of each difficulty type with 0.5. The comparison results show that the predicted probability of the player under the "Medium" difficulty is 0.6, which exceeds the probability threshold. Therefore, the "Medium" difficulty is determined as the target difficulty. At the same time, the predicted probability of the player under the "Hard" difficulty is 0.4, which does not exceed the probability threshold. Therefore, the "Hard" difficulty is not the target difficulty. Since there is only one target difficulty ("Medium" difficulty), the system determines the game difficulty corresponding to the "Medium" difficulty as the predicted game difficulty. If the comparison results show that multiple difficulty types exceed the probability threshold, such as the "Medium" and "Hard" difficulties, the system adds up the game difficulties corresponding to these two difficulties to obtain the predicted game difficulty. Suppose the game difficulty corresponding to the "Medium" difficulty is 5, and the game difficulty corresponding to the "Hard" difficulty is 10. If the player exceeds the probability threshold under both the "Medium" and "Hard" difficulties, then the predicted game difficulty will be 5 + 10 = 15. The game difficulty can be dynamically predicted and adjusted according to the player's performance in the game to ensure that the player can enjoy the game at an appropriate difficulty level. This personalized game experience can not only improve the player's satisfaction but also help developers better understand the player's behavior patterns and further optimize the game design.

[0108] In this way, by determining the predicted game difficulty that the player may face in the game based on the predicted probabilities of difficulties of each difficulty type, and further through the methods of probability threshold comparison and target difficulty identification, the refined and personalized adjustment of the game difficulty can be achieved. The beneficial effects of this processing method are reflected in: It allows the game system to more accurately capture the degree of challenge that the player faces at each difficulty level, and by setting the probability threshold to distinguish which difficulty types are the real challenges that the player faces. This method not only helps to ensure that the player encounters an appropriate amount of challenge in the game, avoiding excessive frustration or lack of challenge, but also can dynamically adjust the game difficulty according to the number of target difficulties, making the game experience more rich and diverse.

[0109] In step 105, when the predicted game difficulty is greater than the difficulty threshold, based on the predicted game difficulty, determine the hint information for giving the player game hints, and display the hint information.

[0110] In some embodiments, when the predicted game difficulty is greater than the difficulty threshold, it indicates that the level of difficulty encountered by the player in the game exceeds the preset expected level. In this case, the system can determine appropriate game hint information according to the predicted game difficulty to help the player overcome the difficulties. The corresponding hint information will be selected or generated from the preset hint library according to the predicted game difficulty. For example, if the predicted game difficulty is hard, the system may choose to provide hints about game strategies or guide the player on how to utilize specific mechanisms in the game.

[0111] As an example, refer to Figure 4 , Figure 4 which is a schematic diagram of the interface of the hint information provided by the embodiments of the present application. When the predicted game difficulty is greater than the difficulty threshold, based on the predicted game difficulty, the hint information 508 for giving game hints to the player is determined and the hint information 508 is displayed.

[0112] In some embodiments, the content of the hint information can be a strategy hint: hinting the player on how to change strategies or take different actions. An operation hint: providing guidance on how to perform a specific operation, such as using a special skill or item. An information hint: providing hints about the game environment or key information of the task to help the player better understand the game state. An incentive hint: giving the player encouraging information to boost the player's confidence and motivation.

[0113] In some embodiments, the trigger timing of the hint information can display the hint information at the following timings: Player request: The hint information is displayed when the player actively requests help. System detection: The hint information is automatically displayed when the system detects that the player may need help in a specific game state. Time interval: The hint information is automatically provided after a specific time interval in the game.

[0114] In some embodiments, the above-mentioned determining the hint information for giving game hints to the player based on the predicted game difficulty can be implemented in the following manner: Obtain the mapping relationship between multiple preset game difficulties and preset hint information, and compare each of the preset game difficulties in the mapping relationship with the predicted game difficulty respectively to obtain the difficulty comparison results of each of the preset game difficulties; when the difficulty comparison result indicates that the preset game difficulty is the same as the predicted game difficulty, determine the preset hint information corresponding to the difficulty comparison result in the mapping relationship as the hint information for giving game hints to the player.

[0115] In some embodiments, developers pre - define different game difficulty levels according to the game design, such as easy, medium, hard, etc. For each preset game difficulty level, developers prepare corresponding hint messages, which are designed to help players overcome challenges at the corresponding difficulty level. Compare the preset game difficulty in the obtained mapping relationship with the predicted game difficulty of the player. The comparison result will indicate which preset game difficulties match the predicted game difficulty. When the comparison result shows that a certain preset game difficulty is the same as the predicted game difficulty, the corresponding preset hint message in the mapping relationship will be used. This ensures that the hint message matches the player's actual game difficulty, improving the pertinence and effectiveness of the hint.

[0116] As an example, in the application scenario of a card game, the following is an illustration of how to use the above - mentioned method to determine the hint message for giving game hints to players: The mapping relationship between preset game difficulty and hint message. Suppose in the Spider Solitaire game, the developer defines the following preset game difficulties and corresponding hint messages: Hint message: Try to clear the left card pile first to increase the operation space. Hint message: Pay attention to the cards on the hidden card pile and plan the card - playing order reasonably. Hint message: It is recommended to use the undo function to backtrack the operation and re - plan the strategy. Assume that the system predicts the player's current game difficulty as medium difficulty based on the player's historical state characteristics and historical operation characteristics. The system will compare the predicted game difficulty with the preset game difficulties: Easy difficulty: Does not match the predicted game difficulty (medium). Medium difficulty: Matches the predicted game difficulty (medium). Hard difficulty: Does not match the predicted game difficulty (medium). Since the predicted game difficulty matches the preset medium difficulty, the system will extract the corresponding hint message from the mapping relationship: Hint message: Pay attention to the cards on the hidden card pile and plan the card - playing order reasonably. The system will display this hint message at an appropriate position in the game interface. For example, display a floating hint box at the bottom of the screen with the content: Pay attention to the cards on the hidden card pile and plan the card - playing order reasonably. Or when the player clicks the "Hint" button, pop up this hint message. If the player still performs poorly at the medium difficulty level, the system can further adjust the hint message. For example: It is recommended to clear the right card pile first to increase the operation space. If the predicted game difficulty is between medium and hard, the hint messages of both difficulties can be combined. For example: Pay attention to the cards on the hidden card pile and use the undo function reasonably.

[0117] In this way, by determining the hint information based on the mapping relationship between the predicted game difficulty and the preset game difficulty, the system can provide accurate and personalized game hints for players, thus significantly enhancing the players' game experience. This method compares the predicted game difficulty with the preset difficulty to ensure that the hint information matches the actual challenge level of the players, avoiding both the decline in the game experience caused by overhints and the player frustration caused by insufficient hints. At the same time, this hint mechanism based on the mapping relationship has high flexibility and scalability, and can dynamically adjust the hint content according to game updates and player feedback, providing a scientific basis for developers to optimize game design.

[0118] In this way, during the process of the player playing the game, the historical state data of the game and the historical operation data triggered by the player for the game are obtained; feature extraction is performed on the historical state data to obtain historical state features, and feature extraction is performed on the historical operation data to obtain historical operation features; based on the historical state features and historical operation features, the game difficulty when the player continues to play the game is predicted to obtain the predicted game difficulty; when the predicted game difficulty is greater than the difficulty threshold, based on the predicted game difficulty, the hint information for giving game hints to the player is determined and the hint information is displayed. In this way, the precise regulation of the game difficulty is achieved by deeply analyzing the player's behavior and game state. In this process, the historical state data of the player in the game and the triggered historical operation data are collected, and these data reflect the player's game performance and preferences. By performing feature extraction on these data, the system can obtain the operation features and state features representing the player's game state, and these features are the basis for predicting the player's game difficulty. Then, these features are used to predict the difficulty that the player may face when continuing to play the game, so as to obtain the predicted game difficulty. When this predicted difficulty exceeds the preset threshold, the corresponding game hint information will be determined and provided according to the prediction result to help the player overcome the challenge. It can not only adapt to the user's needs in real time, but also dynamically adjust the game difficulty according to the player's actual performance to achieve precise regulation.

[0119] Next, an exemplary application of the embodiments of the present application in an application scenario of an actual card game will be described.

[0120] The embodiment of the present application uses advanced machine learning technology to conduct in-depth analysis of the player's game behavior data and game status data, aiming to predict the player's operating habits and the game difficulties that will be encountered, so as to achieve personalized intelligent prompts. Its key technical points include: building a player behavior and game difficulty prediction model based on deep learning. By collecting a large amount of player operation data and game status data, the system uses a neural network algorithm to train the data in an offline environment. To ensure real-time performance, the prediction model is optimized and can run efficiently, and can identify and predict the player's operation mode, strategy selection, and the game difficulties that will be encountered. The intelligent prompt function is integrated into the game interface. When the player plays the game, the system analyzes its current operation, game progress and historical behavior in real time, predicts the difficulties that the player may encounter, provides targeted prompts and suggestions, and enhances the interactive experience of the game. This improved human-computer interaction solution makes the game more in line with the player's personal preferences and actual needs. To protect the privacy of the player, all data collection is anonymized and carried out with the consent of the player. The efficient processing of player data and game status is achieved by using a cloud server and client collaboration. Complex computing tasks are mainly completed in the cloud, and the client deploys an optimized lightweight model to ensure the smooth operation of the game while reducing network transmission delays. The player behavior and game difficulty prediction system can provide customized gaming experience for different players, significantly improving the user satisfaction and market competitiveness of classic card mobile games.

[0121] The embodiment of the present application introduces an intelligent prompt system based on player behavior and game difficulty prediction in the classic card mobile game, which greatly improves the human-computer interaction experience of the game from the product level. The following is a detailed description of the system: Integration of intelligent prompt function, interface design: Add an intelligent prompt area at the appropriate position of the game interface, presented in the form of a translucent floating window to ensure that the key parts of the game are not blocked. The prompt area contains a customizable icon or button, and the player can click to expand or collapse the prompt content. The complete operation process of human-computer interaction: Step 1: Game start. The player opens the game and enters the main interface of the classic card. The newly added intelligent prompt function is presented in the form of a simple icon, which does not affect the initial user experience. Step 2: Real-time data collection. After the player starts the game, the system background collects its operation data and game status data in real time, including the movement of cards, the dwell time, the order of operations, the current game situation (such as the distribution of the deck, movable cards, etc.). All data are collected with the consent of the player and anonymized. Step 3: Behavior and difficulty prediction analysis. The system uses a pre-trained and optimized machine learning model to analyze the player's current operation sequence and game status, and predict the player's possible next operation and upcoming game difficulties.

[0122] In some embodiments, the definition of difficulty: In the present invention, the game difficulty is defined as the situation that may impede the game progress or lead to failure during the game process for the player. For example: Unsolveable situation: The current card combination results in no feasible moves and the game cannot continue. Strategy error: The player selects a sub-optimal or wrong operation, which may lead to the inability to complete the game subsequently. Time pressure: In the time-limited mode, the remaining time for the player is insufficient and they may not be able to complete the game. High error rate: The player makes multiple wrong operations continuously in a short period, indicating possible insufficient understanding of the game strategy. Prediction of difficulty types: The system predicts the specific types that the player may encounter in real time for the above difficulties and evaluates the occurrence probability. Prediction method: Using a deep learning model, inputting the current operation sequence of the player and the game state features, and outputting the prediction result of the difficulty type.

[0123] In some embodiments, Step Four: Provide personalized prompts. When the system predicts that the player may encounter difficulties, the intelligent prompt area will pop up in due course to provide suggestions for the current situation. For example: Prompt feasible operations: When it is detected that an unsolveable situation is about to occur, prompt the player with feasible moves to avoid the unsolveable situation. Strategy suggestions: When the player may make a wrong operation, prompt a better strategy choice. Time reminder: When the time is insufficient, prompt the player to speed up or use overtime items. Step Five: Player selection interaction. The player can choose to view the detailed prompt, ignore the prompt, or adjust the prompt settings. If the player clicks on the prompt, the system will highlight the relevant cards or positions to provide more intuitive guidance. Step Six: Feedback and adaptation. The system dynamically adjusts the frequency and content of the prompt according to the player's feedback on the prompt (such as acceptance, ignoring, adjusting the prompt frequency) to make it more in line with the player's preferences.

[0124] In some embodiments, interface elements and interaction design, customizable settings, prompt frequency adjustment: In the settings menu, the player can choose the frequency of prompt appearance (high, medium, low) or turn off the prompt function completely. Prompt form selection: Provide multiple prompt forms such as text, icons, animations, etc., and the player can choose according to their preferences. Non-intrusive prompt, pop-up method: The prompt pops up with a gentle animation effect and will not suddenly interrupt the player's operation. Position and size: The position and size of the prompt window are optimized to be clearly visible and not interfere with the player's clicking on other interface elements.

[0125] In some embodiments, improved user experience, real-time and accuracy: The intelligent prompt provides the most relevant suggestions based on the player's current game state, operation habits, and upcoming difficulties, avoiding the rigidity and inaccuracy of general prompts. Personalization and adaptation: The system can learn and adapt to the player's operation style, making the prompt content more and more in line with the player's needs and enhancing the personalized experience of the game.

[0126] In this way, the operation efficiency is improved: through intelligent prompts, players can find the best operation path more quickly, avoiding getting into unsolvable situations or making strategic mistakes. The game stickiness is enhanced: personalized interactions improve players' satisfaction and game stickiness, encouraging players to continuously participate in the game. The sense of frustration is reduced: by predicting and prompting players in advance about the difficulties they will encounter, the sense of frustration caused by failure is reduced, improving the entertainment of the game.

[0127] In some embodiments, there is a no-solution warning: when the system detects that the player's current operation may lead to an unsolvable situation, it promptly prompts the player to adjust the strategy. Strategy optimization: for experienced players, the system can identify their common strategies and provide optimization suggestions at critical moments, enhancing the game challenge. Time management: in the time-limited mode, the system reminds the player to pay attention to the remaining time or suggests using items to extend the time.

[0128] The embodiments of this application mainly involve a player behavior and game difficulty prediction system based on machine learning at the technical level, including links such as data collection, preprocessing, model training, real-time prediction, and prompt generation. The following elaborates on this technical solution in detail.

[0129] In some embodiments, the system mainly consists of the following modules: Data collection module: responsible for collecting the player's game operation data and game state data. Data preprocessing module: cleans and formats the collected data. Behavior and difficulty prediction model: based on deep learning algorithms, predicts the player's operation habits and upcoming game difficulties. Prompt generation module: generates personalized prompts according to the prediction results. Client-server interaction module: realizes the upload of data and the issuance of prompts.

[0130] In some embodiments, the data processing flow: Step 1: Data collection, real-time data acquisition: when the player is performing game operations, the client records the operation data and game state data in real time, including: Operation data: click position, operation sequence, time interval, moved cards, and target position, etc. Game state data: the distribution of the current card stack, remaining available cards, set of feasible operations, etc.

[0131] In some embodiments, data caching and upload: to reduce the network burden and protect data security, the client temporarily stores the data and uploads it to the server in a suitable network environment. Step 2: Data preprocessing. Data cleaning: the server side cleans the received data, removing abnormal and invalid data. Feature extraction: extracts key features from the original data, including: Operation sequence features: temporal features, frequency features, common operation patterns, etc. of operations. Game state features: card stack state, number of movable cards, remaining possibilities, etc. Label generation: generates labels for the training data according to the game rules and results, such as whether the operation leads to no solution, whether it is a strategic mistake, etc.

[0132] In some embodiments, Step 3: Model Training, Algorithm Selection: Select a deep learning model suitable for sequence data and state prediction, such as a hybrid model combining a Convolutional Neural Network (CNN) and a Long Short-Term Memory Network (LSTM). Training Process: Use a large amount of historical data to perform offline training on the model on a cloud server, and adjust the parameters to improve the prediction accuracy. Model Optimization: Through techniques such as model pruning, quantization, and knowledge distillation, obtain a lightweight prediction model suitable for real-time operation on the client side. Step 4: Real-time Prediction and Hint Generation, Real-time Data Input: During the game process, the client inputs the player's current operation sequence and game state into the locally deployed prediction model. Prediction Calculation: The client uses the optimized model to predict in real-time the player's possible next operations and upcoming game difficulties. Difficulty Type Prediction: The model outputs the possible difficulty types and their occurrence probabilities, such as the risk of an unsolvable situation, the possibility of a strategic error, etc. Hint Generation: Based on the prediction results, generate personalized hint information to help the player avoid difficulties or provide solutions.

[0133] In some embodiments, Step 5: Hint Display. Hint Display: The client displays the hint in an appropriate manner in the game interface without interfering with the player's normal operations.

[0134] In some embodiments, Data Collection and Preprocessing. Data Format: The operation data and game state data are indexed by timestamp and contain detailed operation and state information. Data Compression: Adopt a lightweight compression algorithm to reduce the data transmission volume. Preprocessing Method: Use methods such as sliding window and sequence truncation to convert the operation sequence and state sequence into an input format acceptable to the model.

[0135] In some embodiments, Model Training and Optimization. Model Architecture: Use a deep learning model combining CNN and LSTM. CNN is used to extract the spatial features of the game state, and LSTM is used to capture the temporal features of the operation sequence. Loss Function Design: Design a suitable loss function for difficulty prediction, considering multi-task learning, predicting both the player's behavior and the difficulty type. Model Optimization: Through techniques such as model pruning, quantization, and knowledge distillation, obtain a model suitable for running on the client side.

[0136] In some embodiments, Real-time Prediction and Hints. Efficient Computation: The client model is optimized to run efficiently on mobile devices, and the prediction latency is controlled within milliseconds. Resource Management: Reasonably allocate the computing resources of the device and adopt techniques such as asynchronous computing to ensure that the prediction process does not affect the smoothness of the game.

[0137] In this way, the prediction accuracy is improved: By introducing advanced deep learning models, the limitations of traditional statistical methods are overcome, and complex operation patterns and game states can be captured to accurately predict the difficulties that players will encounter. Real-time performance is achieved: Through model optimization and deployment on the client side, it is ensured that the prediction and hint generation are completed within an acceptable time without affecting the players' gaming experience. Personalized hints: The model makes predictions based on each player's unique operation data and game progress, providing highly customized hints. Reducing the burden on the client: The model is optimized to consume less resources, ensuring the smooth operation of the game and the stability of the device.

[0138] In this way, the personalized experience of the game is enhanced: By using machine learning technology to deeply analyze players' operation habits and upcoming game difficulties, the system can provide targeted intelligent hints. Each player can enjoy tailor-made game guidance, enhancing the attractiveness and user stickiness of the game. The operation efficiency and game winning rate are improved: The intelligent hint function helps players quickly find the best operation path, avoiding getting into insoluble situations or making strategic mistakes, improving the game clearance efficiency and enhancing the players' sense of achievement. Reducing the difficulty for novice players to get started: For novice players, the real-time guidance and difficulty warning provided by the system can effectively reduce the learning curve, helping them become familiar with the game rules and operation methods faster, expanding the user group of the game. Enhancing the interactivity and immersion of the game: The personalized hints make the interaction between players and the game closer, enhancing the immersive experience of the game and increasing the entertainment and fun of the game. Reducing the burden on mobile devices: By deploying an optimized lightweight model on the client side, it is ensured that the game runs smoothly and stably without affecting the device performance and battery life. To sum up, this technical solution not only effectively solves the problems in the prior art such as low prediction accuracy, poor real-time performance, and lack of prediction and hints for the difficulties that players will encounter, but also brings a better gaming experience to players, which is of great significance for improving the quality of game products and user satisfaction.

[0139] In some embodiments, for the above technical solutions, the following alternative solutions can be considered: Algorithm substitution: In addition to using deep learning models such as CNN and LSTM for player behavior and difficulty prediction, graph neural networks (GNNs) can also be tried to model the deck relationships in the game state to enhance the understanding of complex game states. Reinforcement learning: Adopt the Reinforcement Learning algorithm to train an intelligent agent to simulate player operations, predict the best operations and possible difficulty points through a policy network, and directly provide more effective guidance for players. Federated learning: Adopt the method of Federated Learning. The model is locally trained on the players' devices, and the server only aggregates the model parameters, which not only improves data security but also enriches the diversity of the model. Diversified hint forms: In addition to providing text and icon hints in the game interface, various interaction methods such as voice hints and vibration feedback can also be adopted. For example, when the player wears headphones, the system can use a voice assistant to broadcast hint information in real time to enhance the immersion of the game. Edge computing: In order to further reduce latency and improve efficiency, edge computing technology can be adopted. Deploy the prediction model on an edge server closer to the player to shorten the data transmission path and improve real-time performance. Model lightweight technology: Adopt model lightweight technologies such as model pruning, quantization, and knowledge distillation to further reduce the size and computational complexity of the model to adapt to more types of mobile devices. Data augmentation technology: Use data augmentation technology to generate more simulated player operation data, enrich the training data of the model, and improve the generalization ability of the model.

[0140] It can be understood that in the embodiments of the present application, relevant data such as historical state data and historical operation data are involved. When the embodiments of the present application are applied to specific products or technologies, user permission or consent needs to be obtained, and the collection, use, and processing of relevant data need to comply with the relevant laws, regulations, and standards of relevant countries and regions.

[0141] Next, the exemplary structure of the game processing device 455 provided in the embodiments of the present application implemented as software modules will be continued. In some embodiments, such as Figure 2As shown, the software modules in the game processing device 455 stored in the memory 450 may include: an acquisition module, configured to acquire historical state data of the game and historical operation data triggered by the player for the game during the process of the player playing the game; a feature extraction module, configured to extract features from the historical state data to obtain historical state features and extract features from the historical operation data to obtain historical operation features; a prediction module, configured to predict the game difficulty when the player continues to play the game based on the historical state features and the historical operation features to obtain a predicted game difficulty; a display module, configured to, when the predicted game difficulty is greater than a difficulty threshold, determine prompt information for prompting the player for the game based on the predicted game difficulty and display the prompt information.

[0142] In some embodiments, the above acquisition module is further configured to, during the process of the player playing the game, in response to an interaction operation for the game, acquire the interaction duration between the game and the player, compare the interaction duration with an interaction duration threshold to obtain a duration comparison result; when the duration comparison result indicates that the interaction duration is greater than or equal to the interaction duration threshold, send an acquisition request for the historical state data and the historical operation data to the server; and receive the historical state data and the historical operation data queried by the server for the acquisition request.

[0143] In some embodiments, the above acquisition module is further configured to, during the process of the player playing the game, respectively perform the following processing at each preset moment: collect historical state data and historical operation data before the preset moment, perform data cleaning on the historical state data and historical operation data corresponding to the preset moment to obtain cleaned historical state data and historical operation data; and send the cleaned historical state data and historical operation data to the server, so that the server queries the historical state data and the historical operation data when receiving the acquisition request.

[0144] In some embodiments, the above acquisition module is further configured to, during the process of the player playing the game, respectively perform the following processing at each preset moment: collect historical state data and historical operation data before the preset moment, perform data cleaning on the historical state data and historical operation data corresponding to the preset moment to obtain cleaned historical state data and historical operation data; and send the cleaned historical state data and historical operation data to the server, so that the server queries the historical state data and the historical operation data when receiving the acquisition request.

[0145] In some embodiments, the above-mentioned acquisition module is further configured to, during the process of the player playing the game, in response to an interaction operation for the game, acquire the historical acquisition times of the historical state data and the historical operation data; when the historical acquisition times are zero, acquire the historical state data and the historical operation data; when the historical acquisition times are at least one, acquire the historical acquisition moment of the most recent historical state data and the historical operation data, and determine the time difference between the historical acquisition moment and the current moment; when the time difference is greater than or equal to the time difference threshold, acquire the historical state data and the historical operation data.

[0146] In some embodiments, the above-mentioned prediction module is further configured to perform a difficulty prediction on the player based on the historical state features and the historical operation features, to obtain the prediction probabilities of the player encountering difficulties of each difficulty type; based on the prediction probabilities of the difficulties of each difficulty type, determine the predicted game difficulty when the player continues to play the game.

[0147] In some embodiments, the above-mentioned difficulty prediction is implemented through a difficulty prediction model. The difficulty prediction model includes a feature fusion layer and a difficulty prediction layer. The above-mentioned prediction module is further configured to call the feature fusion layer to perform feature fusion on the historical state features and the historical operation features to obtain fused features; call the difficulty prediction layer to perform a difficulty prediction on the player based on the fused features to obtain the prediction probabilities of the player encountering each of the difficulty types.

[0148] In some embodiments, the above-mentioned prediction module is further configured to acquire historical state feature samples and historical operation feature samples, and call an initial difficulty prediction model to perform a difficulty prediction on player samples corresponding to the historical state feature samples and the historical operation feature samples based on the historical state feature samples and the historical operation feature samples, to obtain the sample prediction probabilities of the player samples encountering difficulties of each of the difficulty types; for each of the difficulty types, determine the loss value corresponding to the difficulty type based on the sample prediction probability of the difficulty type and the label prediction probability of the difficulty type; based on each of the loss values, train the initial difficulty prediction model to obtain the difficulty prediction model.

[0149] In some embodiments, the above-mentioned prediction module is further configured to compare the prediction probabilities of the difficulties of each of the difficulty types with a probability threshold respectively to obtain the probability comparison results of each of the difficulties; determine the difficulties whose probability comparison results indicate that the prediction probabilities are greater than the probability threshold as target difficulties; when the number of the target difficulties is one, determine the game difficulty corresponding to the target difficulty as the predicted game difficulty; when the number of the target difficulties is multiple, determine the sum of the game difficulties corresponding to each of the target difficulties as the predicted game difficulty.

[0150] In some embodiments, the display module is further configured to obtain a mapping relationship between a plurality of preset game difficulties and preset hint messages, compare each of the preset game difficulties in the mapping relationship with the predicted game difficulty respectively, and obtain a difficulty comparison result for each of the preset game difficulties; when the difficulty comparison result indicates that the preset game difficulty is the same as the predicted game difficulty, determine the preset hint message corresponding to the difficulty comparison result in the mapping relationship as the hint message for giving a game hint to the player.

[0151] The embodiment of the present application provides a computer program product, which includes a computer program or computer-executable instructions, and the computer program or computer-executable instructions are stored in a computer-readable storage medium. The processor of the electronic device reads the computer-executable instructions from the computer-readable storage medium, and the processor executes the computer-executable instructions, so that the electronic device executes the game processing method described above in the embodiment of the present application.

[0152] The embodiment of the present application provides a computer-readable storage medium storing computer-executable instructions, where the computer-executable instructions are stored, and when the computer-executable instructions are executed by a processor, the processor will be caused to execute the game processing method provided by the embodiment of the present application. For example, Figure 3 the game processing method shown.

[0153] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disc, or CD-ROM; it may also be various electronic devices including one or any combination of the above memories.

[0154] In some embodiments, the computer-executable instructions may be in the form of a program, software, software module, script, or code, and may be 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 being deployed as an independent program or being deployed as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0155] As an example, the computer-executable instructions may or may not correspond to a file in the file system, and may be stored as part of a file that stores other programs or data. For example, they may be stored in one or more scripts in a HyperText Markup Language (HTML) document, stored in a single file dedicated to the program being discussed, or stored in multiple cooperating files (for example, files that store one or more modules, subroutines, or code portions).

[0156] As an example, the computer-executable instructions can be deployed to execute on one electronic device, or on multiple electronic devices located at one location, or on multiple electronic devices distributed at multiple locations and interconnected through a communication network.

[0157] In summary, the embodiments of the present application have the following beneficial effects:

[0158] (1) During the process of a player playing a game, obtain the historical state data of the game and the historical operation data triggered by the player for the game; extract features from the historical state data to obtain historical state features, and extract features from the historical operation data to obtain historical operation features; based on the historical state features and historical operation features, predict the game difficulty when the player continues to play the game to obtain the predicted game difficulty; when the predicted game difficulty is greater than the difficulty threshold, based on the predicted game difficulty, determine the prompt information for prompting the player during the game and display the prompt information. In this way, the precise control of the game difficulty is achieved by deeply analyzing the player's behavior and game state. During this process, collect the historical state data and the triggered historical operation data of the player in the game, and these data reflect the player's game performance and preferences. By extracting features from these data, the system can obtain the operation features and state features representing the player's game state, and these features are the basis for predicting the player's game difficulty. Then, use these features to predict the difficulty that the player may face when continuing to play the game, so as to obtain the predicted game difficulty. When this predicted difficulty exceeds the preset threshold, the corresponding game prompt information will be determined and provided according to the prediction result to help the player overcome the challenge. It can not only adapt to the user's needs in real time, but also dynamically adjust the game difficulty according to the player's actual performance to achieve precise control.

[0159] (2) During the process of a player playing a game, by detecting the interaction duration between the player and the game and comparing it with the preset interaction duration threshold, a request for obtaining the historical state data and historical operation data can be automatically triggered after the player has invested sufficient time. This mechanism can effectively identify the player's participation and potential difficulties in the game, so as to provide data support for subsequent analysis and optimization. For example, when the player stays at a certain level for too long, prompt information can be generated based on historical data or the game difficulty can be adjusted to help the player pass the level smoothly. This way of obtaining data on demand not only improves the efficiency of the system, but also provides a more personalized and smooth game experience for the player, and at the same time provides a scientific basis for developers to optimize the game design, ultimately improving the user satisfaction and stickiness of the game.

[0160] (3) By regularly collecting, cleaning, and uploading historical status data and historical operation data at preset times during the player's game process, the system can ensure the accuracy, integrity, and availability of the data. Data cleaning removes invalid and incorrect information, improving data quality, while the pre-storage mechanism significantly enhances the efficiency and response speed of server queries. This preprocessing process provides reliable data support for subsequent difficulty prediction, hint generation, and personalized game experience optimization, while reducing the burden of real-time data processing. Ultimately, this mechanism not only enhances the player's game experience but also provides a scientific basis for developers to optimize game design, enhance user stickiness and satisfaction.

[0161] (4) By dynamically determining the acquisition timing of historical status data and historical operation data during the player's game process, the system can effectively avoid duplicate data acquisition, reduce resource waste, and ensure data timeliness. When the historical acquisition count is zero, the system immediately acquires data to meet the initial requirements; when the historical acquisition count is at least one, the system determines whether to re-acquire data by comparing the time difference between the most recent acquisition time and the current time. This mechanism not only improves the system's operating efficiency but also provides accurate and timely data support for subsequent intelligent functions such as difficulty adjustment and hint generation, thereby enhancing the player's game experience and the overall performance of the system.

[0162] (5) It can more accurately predict the player's game performance and potential difficulties, thus providing a scientific basis for dynamically adjusting the game difficulty and generating personalized hint information. This prediction method based on data and models not only enhances the player's game experience but also helps developers understand the player behavior pattern, further optimize game design, and enhance the game's attractiveness and user stickiness. Based on the probability distribution of difficulty prediction, the system can implement more flexible and refined game difficulty adjustment strategies to ensure that the game can challenge the player's ability while avoiding overly frustrating the player's enthusiasm, thereby maintaining the game's fun while increasing the player's satisfaction and retention rate. Through this mechanism, the game can better adapt to the player's needs, achieve a positive interaction between the player and the game, and promote the sustainable development and innovation of the game industry.

[0163] (6) The difficult prediction model that adopts the feature fusion layer and the difficult prediction layer can significantly improve the accuracy and efficiency of prediction by fusing historical state features and historical operation features and then making difficult predictions. It can not only better capture the complex relationship between the player's behavior and the game environment, but also provide strong support for the game's dynamic difficulty adjustment and personalized experience. The feature fusion layer effectively integrates data from different sources, enabling the model to comprehensively understand the player's game performance from multiple dimensions, while the difficult prediction layer accurately predicts the difficulties that the player may encounter based on these fused features, so that the game can respond to the player's needs in a timely manner and provide a more considerate game experience. It helps to improve the player's game satisfaction, increase user stickiness, and at the same time provides a data basis for game developers to deeply understand the player's behavior, provides a direction for game optimization and iteration, and promotes the innovative development of the game industry.

[0164] (7) By obtaining historical state feature samples and historical operation feature samples and using these samples to train the initial difficult prediction model, the prediction accuracy and generalization ability of the model can be significantly improved. This method enables the model to better understand and learn the complex relationship between the player's behavior pattern and the game environment, and thus be more accurate in predicting the possibility of the player encountering difficulties. In addition, by calculating the loss value between the sample prediction probability and the label prediction probability and training the model specifically, the model parameters can be effectively optimized and the prediction error can be reduced.

[0165] (8) By determining the predicted game difficulty that the player may face in the game based on the prediction probability of the difficulties of each difficulty type, and further through the methods of probability threshold comparison and target difficulty identification, the refined and personalized adjustment of the game difficulty can be achieved. The beneficial effects of this processing method are reflected in that it allows the game system to more accurately capture the degree of challenge that the player faces at each difficulty level, and distinguishes which difficulty types are the real challenges faced by the player by setting a probability threshold. This method not only helps to ensure that the player encounters an appropriate amount of challenge in the game, avoiding excessive frustration or lack of challenge, but also can dynamically adjust the game difficulty according to the number of target difficulties, making the game experience more rich and diverse.

[0166] (9) By determining the prompt information based on the mapping relationship between the predicted game difficulty and the preset game difficulty, the system can provide accurate and personalized game prompts for players, thus significantly enhancing the players' gaming experience. This method compares the predicted game difficulty with the preset difficulty to ensure that the prompt information matches the players' actual challenge level, avoiding both the decline in the gaming experience caused by overprompting and the players' frustration caused by insufficient prompting. At the same time, this prompt mechanism based on the mapping relationship has high flexibility and scalability, and can dynamically adjust the prompt content according to game updates and player feedback, providing a scientific basis for developers to optimize game design.

[0167] (10) Improve operation efficiency: Through intelligent prompts, players can find the best operation path more quickly, avoiding getting into unsolvable situations or making strategic mistakes. Enhance game stickiness: Personalized interactions increase players' satisfaction and game stickiness, encouraging players to continuously participate in the game. Reduce frustration: By predicting and prompting players about the upcoming difficulties in advance, the frustration caused by failure is reduced, enhancing the entertainment value of the game.

[0168] (11) Unsolvable warning: When the system detects that the player's current operation may lead to an unsolvable situation, it promptly prompts the player to adjust the strategy. Strategy optimization: For experienced players, the system can identify their common strategies and provide optimization suggestions at critical moments, enhancing the game challenge. Time management: In the time-limited mode, the system reminds the player to pay attention to the remaining time or suggests using items to extend the time.

[0169] (12) Improve prediction accuracy: By introducing advanced deep learning models, the limitations of traditional statistical methods are overcome, enabling the capture of complex operation patterns and game states and accurately predicting the difficulties that players will encounter. Achieve real-time performance: Through model optimization and deployment on the client side, it is ensured that the generation of predictions and prompts is completed within an acceptable time without affecting the players' gaming experience. Personalized prompts: The model makes predictions based on each player's unique operation data and game progress, providing highly customized prompts. Reduce the client burden: The model is optimized to consume less resources, ensuring the smooth operation of the game and the stability of the device.

[0170] (13) Enhance the personalized gaming experience: By leveraging machine learning technologies to deeply analyze players' operation habits and upcoming gaming difficulties, the system can provide targeted intelligent tips. Each player can enjoy customized gaming guidance, enhancing the game's attractiveness and user stickiness. Improve operation efficiency and gaming win rate: The intelligent tip function helps players quickly find the optimal operation path, avoid getting into insoluble situations or making strategic mistakes, improve the game clearance efficiency, and enhance players' sense of achievement. Reduce the learning curve for novice players: For novice players, the real-time guidance and difficulty warning provided by the system can effectively reduce the learning curve, help them quickly familiarize themselves with the game rules and operation methods, and expand the game's user base. Enhance the interactivity and immersion of the game: Personalized tips make the interaction between players and the game closer, enhance the immersive experience of the game, and increase the entertainment and fun of the game. Alleviate the burden on mobile devices: By deploying an optimized lightweight model on the client side, ensure the smoothness and stability of the game operation without affecting the device performance and battery life. In summary, this technical solution not only effectively solves the problems of low prediction accuracy, poor real-time performance, lack of prediction and tips for upcoming difficulties of players in the prior art, but also brings a better gaming experience to players, which is of great significance to improving the quality of game products and user satisfaction.

[0171] The above is only an example of this application and is not intended to limit the protection scope of this application. Any modifications, equivalent replacements, and improvements made within the spirit and scope of this application are included in the protection scope of this application.

Claims

1. A method for processing a game, characterized in that: The method comprises: During the process of the player playing the game, obtaining historical status data of the game and historical operation data triggered by the player for the game; Performing feature extraction on the historical state data to obtain historical state features, and performing feature extraction on the historical operation data to obtain historical operation features; Based on the historical state characteristics and the historical operation characteristics, predicting the game difficulty when the player continues to play the game, to obtain a predicted game difficulty; When the predicted game difficulty is greater than a difficulty threshold, prompt information for providing game prompts to the player is determined based on the predicted game difficulty, and the prompt information is displayed.

2. The method according to claim 1, characterized in that The process of obtaining the historical state data of the game and the historical operation data triggered by the player for the game during the player playing the game includes: In the process of the player playing the game, in response to an interactive operation on the game, obtaining an interaction time between the game and the player, and comparing the interaction time with an interaction time threshold to obtain a time comparison result; When the duration comparison result indicates that the interaction duration is greater than or equal to the interaction duration threshold, sending a request for obtaining the historical state data and the historical operation data to a server; The historical status data and the historical operation data obtained by the server in response to the acquisition request query are received.

3. The method according to claim 2, characterized in that Before obtaining the historical status data of the game and the historical operation data triggered by the player on the game, the method further includes: When the player plays the game, the following processing is performed at each preset time: Collecting historical status data and historical operation data before the preset time, performing data cleaning on the historical status data and historical operation data corresponding to the preset time, and obtaining cleaned historical status data and historical operation data; The cleaned historical status data and historical operation data are sent to the server, so that the server queries the historical status data and the historical operation data when receiving the acquisition request.

4. The method according to claim 1, characterized in that: The process of obtaining the historical state data of the game and the historical operation data triggered by the player for the game during the player playing the game includes: In the process of the player playing the game, in response to an interactive operation on the game, acquiring the historical state data and the historical acquisition times of the historical operation data; When the number of history acquisition times is zero, acquiring the history status data and the history operation data; When the number of history acquisition times is at least one, acquiring the most recent history acquisition time of the history status data and the history operation data, and determining the time difference between the history acquisition time and the current time; When the time difference is greater than or equal to a time difference threshold, the historical state data and the historical operation data are acquired.

5. The method according to claim 1, characterized in that The predicting the game difficulty when the player continues to play the game based on the historical state feature and the historical operation feature to obtain the predicted game difficulty includes: Based on the historical state characteristics and the historical operation characteristics, predict the difficulty of the player to obtain the predicted probability of the player encountering difficulties of each difficulty type; Based on the predicted probability of difficulty of each of the difficulty types, a predicted game difficulty when the player continues to play the game is determined.

6. The method according to claim 5, characterized in that The difficulty prediction is implemented by a difficulty prediction model, which includes a feature fusion layer and a difficulty prediction layer. The difficulty prediction is performed on the player based on the historical state features and the historical operation features to obtain the predicted probability of the player encountering difficulties of each difficulty type, including: Calling the feature fusion layer to perform feature fusion on the historical state feature and the historical operation feature to obtain a fusion feature; The difficulty prediction layer is called to perform difficulty prediction on the player based on the fusion features to obtain the predicted probability that the player will encounter each difficulty type.

7. The method according to claim 6, characterized in that Before performing difficulty prediction for the player based on the historical state characteristics and the historical operation characteristics to obtain the predicted probability of the player encountering difficulties of each difficulty type, the method further includes: Obtaining historical state feature samples and historical operation feature samples, and calling an initial difficulty prediction model, based on the historical state feature samples and the historical operation feature samples, performing difficulty prediction on player samples corresponding to the historical state feature samples and the historical operation feature samples, and obtaining sample prediction probabilities that the player samples will encounter difficulties of each difficulty type; For each difficulty type, based on the sample prediction probability of the difficulty type and the label prediction probability of the difficulty type, determine the loss value corresponding to the difficulty type; Based on each of the loss values, the initial difficulty prediction model is trained to obtain the difficulty prediction model.

8. The method according to claim 5, characterized in that The step of determining the predicted game difficulty when the player continues to play the game based on the predicted probability of difficulty of each difficulty type includes: Compare the predicted probability of difficulty of each difficulty type with the probability threshold respectively to obtain the probability comparison result of each difficulty; Determine the difficulty for which the probability comparison result indicates that the predicted probability is greater than the probability threshold as a target difficulty; When the number of the target difficulties is one, the game difficulty corresponding to the target difficulty is determined as the predicted game difficulty; When there are multiple target difficulties, the sum of the game difficulties corresponding to the target difficulties is determined as the predicted game difficulty.

9. The method according to claim 1, characterized in that: The determining, based on the predicted game difficulty, prompt information for providing game prompts to the player includes: Acquire a mapping relationship between a plurality of preset game difficulties and preset prompt information, and compare each of the preset game difficulties in the mapping relationship with the predicted game difficulty to obtain a difficulty comparison result of each of the preset game difficulties; When the difficulty comparison result indicates that the preset game difficulty is the same as the predicted game difficulty, the preset prompt information corresponding to the difficulty comparison result in the mapping relationship is determined as prompt information for providing game prompts to the player.

10. A game processing device, characterized in that: The device comprises: An acquisition module, used for acquiring historical state data of the game and historical operation data triggered by the player for the game during the process of the player playing the game; A feature extraction module, used to extract features from the historical state data to obtain historical state features, and to extract features from the historical operation data to obtain historical operation features; A prediction module, configured to predict the game difficulty of the player when the player continues to play the game based on the historical state characteristics and the historical operation characteristics, to obtain a predicted game difficulty; The display module is used to determine prompt information for providing game prompts to the player based on the predicted game difficulty when the predicted game difficulty is greater than a difficulty threshold, and display the prompt information.

Images