Training strategy processing method, device and equipment, computer readable storage medium and computer program product

Abstract

The invention provides a training strategy processing method and device, electronic equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: in response to a training strategy customization operation for a first virtual object in a virtual scene corresponding to competitive sports, based on behavior data of competitive interaction of the first virtual object in the virtual scene, determining a first reference object matched with the first virtual object; and according to the first reference object, obtaining a first training strategy adaptive to the first virtual object, and outputting the first training strategy. According to the invention, the pertinence of the training strategy can be improved.

Description

Technical Field

[0001] The present application relates to computer technology, and in particular, to a method, apparatus, device, storage medium, and program product for processing training strategies. Background Art

[0002] In the application of virtual scenarios (such as game scenarios), the behavior of game characters can be improved to adapt to different gameplay styles. However, the behavior of game characters usually operates within preset rules and predefined logics, and its decision-making process may not be able to truly reflect the complex and dynamically changing situations in the real world, such as the on-the-spot contingency strategies in live competitions, resulting in the lack of real-time and targeted nature of this way of improving the behavior of game characters. Summary of the Invention

[0003] Embodiments of the present application provide a method, apparatus, electronic device, computer-readable storage medium, and computer program product for processing training strategies, which can improve the pertinence of the output training strategies.

[0004] The technical solution of the embodiments of the present application is implemented as follows:

[0005] Embodiments of the present application provide a method for processing training strategies, including:

[0006] In response to a training strategy customization operation for a first virtual object in a virtual scenario corresponding to competitive sports, based on the behavior data of the first virtual object in the competitive interaction in the virtual scenario, determine a first reference object that matches the first virtual object;

[0007] According to the first reference object, obtain a first training strategy adapted to the first virtual object, and output the first training strategy.

[0008] Embodiments of the present application provide a device for processing training strategies, including:

[0009] A reference determination module, configured to, in response to a strategy customization operation for a first virtual object in a virtual scenario corresponding to competitive sports, based on the behavior data of the first virtual object in the competitive interaction in the virtual scenario, determine a first reference object that matches the first virtual object;

[0010] A strategy processing module, configured to, based on the first reference object, generate a first training strategy adapted to the first virtual object, and output the first training strategy.

[0011] In the above solution, the reference determination module is further configured to extract features from the behavior data of the first virtual object in the virtual scene for competitive interaction, so as to obtain behavior features characterizing the competitive style of the first virtual object; match the behavior features with the key features of multiple reference objects, and determine a first reference object that matches the first virtual object from the multiple reference objects based on the matching result.

[0012] In the above solution, when the behavior data of the first virtual object in the virtual scene for competitive interaction includes behavior data of multiple competitive actions, the reference determination module is further configured to perform normalization processing on the behavior data to obtain a behavior data sequence; encode the behavior data sequence based on the time dependence of each competitive action in the behavior data sequence to obtain a feature vector of the behavior data sequence, where the feature vector indicates at least one of the following action indicators of each competitive action: action type, action frequency, action coherence, action success rate; evaluate the competitive style of the first virtual object based on the feature vector, and determine behavior features characterizing the competitive style of the first virtual object based on the evaluation result.

[0013] In the above solution, the reference determination module is further configured to evaluate the competitive style of the first virtual object based on the competitive evaluation rules corresponding to the virtual scene and the feature vector to obtain an evaluation result of the competitive style corresponding to the feature vector; determine a standard feature corresponding to the competitive style of the feature vector, and use the standard feature as the behavior feature characterizing the competitive style of the first virtual object.

[0014] In the above solution, the reference determination module is further configured to calculate the similarity between the behavior features and the key features of each reference object, and use the similarity between the behavior features and the key features of each reference object as the matching result; screen out the reference object corresponding to the maximum similarity from the multiple reference objects, and use the screened reference object as the first reference object that matches the first virtual object.

[0015] In the above solution, the strategy processing module is further configured to generate a reference training strategy for the first reference object based on the competitive style of the first reference object, and use the reference training strategy as the first training strategy adapted to the first virtual object.

[0016] In the above solution, the policy processing module is further configured to determine the competitive disadvantage of the first virtual object based on the behavior characteristics of the first virtual object, and generate a strengthening training policy for overcoming the competitive disadvantage based on the competitive disadvantage; determine the reference training policy of the first reference object, and fuse the reference training policy and the strengthening training policy to obtain the first training policy adapted to the first virtual object.

[0017] In the above solution, the training policy customization operation is triggered by triggering a policy customization entry, and the device further includes: an entry display module, configured to perform at least one of the following: display the policy customization entry when the competitive interaction of the first virtual object ends; display the policy customization entry in the cultivation training interface of the first virtual object; display the policy customization entry during the competitive interaction of the first virtual object.

[0018] In the above solution, after determining the first reference object that matches the first virtual object, the device further includes: a competitive style display module, configured to display an analysis interface of the competitive style of the first virtual object based on the first reference object; display the competitive style of the first virtual object, the first reference object that matches the first virtual object, the competitive advantages and disadvantages of the first reference object, and training tips for overcoming the competitive disadvantages in the competitive style analysis interface.

[0019] In the above solution, the policy processing module is further configured to display a training tip interface of the first virtual object; in response to the first training policy including sub-training policies for multiple training items, display training entries for each of the training items in the training tip interface; and output the sub-training policy of the training item corresponding to the selected training entry in response to a selection operation.

[0020] In the above solution, the policy processing module is further configured to determine an influencing factor for influencing the training priority of the training item, where the influencing factor includes at least one of the following: required duration, difficulty, type, and synergistic effect on other training items; predict the training priorities of the training items based on the influencing factor, and display the training entries for the training items in the training tip interface in a manner corresponding to the training priorities.

[0021] In the above solution, the policy processing module is further configured to output the first training policy in the form of policy prompt information in response to meeting the policy prompt condition; where the policy prompt information includes a target competitive action recommended to be executed based on the first training policy and the timing of executing the target competitive action.

[0022] In the above solution, the device further includes: a condition determination module, configured to determine that the policy prompt condition is met when a policy prompt entry is displayed in the interface of the virtual scene and in response to a trigger operation on the policy prompt entry; or, determine that the policy prompt condition is met in response to the interaction progress of the first virtual object in the competitive interaction reaching a target progress, where the target progress includes at least one of the following: there is an association between the competitive action currently performed by the first virtual object and the target competitive action recommended to be performed by the first training policy, and the competitive disadvantage targeted by the first training policy appears on the first virtual object.

[0023] In the above solution, after outputting the first training policy, the device further includes: a policy update module, configured to determine the improvement degree of each execution of the first training policy for the competitive disadvantage when the first virtual object executes the first training policy multiple times to overcome the competitive disadvantage; fuse the improvement degrees of each execution of the first training policy for the competitive disadvantage to obtain a comprehensive improvement degree; and in response to the comprehensive improvement degree being lower than a degree threshold, replace the first training policy with a second training policy, where the second training policy is generated based on the skill style of a second reference object, and the similarity between the behavioral characteristics of the first virtual object and the key characteristics of the second reference object is the same as the similarity between the behavioral characteristics of the first virtual object and the key characteristics of the first reference object.

[0024] In the above solution, after outputting the first training policy, the policy update module is further configured to control the first virtual object to execute the first training policy in the competitive interaction in response to a usage operation on the first training policy; detect the competitive style of the first virtual object based on new behavioral data during the process of the first virtual object executing the first training policy; and adjust the first training policy when there is a difference between the current competitive style and the original competitive style, so that the adjusted first training policy is adapted to the current competitive style.

[0025] In the above solution, after outputting the first training policy, the policy update module is further configured to obtain feedback information on the first training policy in response to a feedback operation; and update and adjust the first training policy based on the feedback information.

[0026] In the above solution, after outputting the first training strategy, the device further includes: a strategy comparison module, configured to create a training challenge task for the first training strategy, and send an invitation request for the training challenge task to the terminal of a second virtual object, where the second virtual object matches the first target object; in response to the second virtual object agreeing to the invitation request and joining the training challenge task, comparing the training result of the first virtual object for the first training strategy with the training result of the second virtual object for the first training strategy.

[0027] An embodiment of the present application provides an electronic device, including:

[0028] a memory, configured to store computer-executable instructions or a computer program;

[0029] a processor, configured to implement the training strategy processing method provided by the embodiment of the present application when executing the computer-executable instructions or the computer program stored in the memory.

[0030] An embodiment of the present application provides a computer-readable storage medium, storing computer-executable instructions or a computer program, configured to implement the training strategy processing method provided by the embodiment of the present application when being executed by a processor.

[0031] An embodiment of the present application provides a computer program product, including a computer program or computer-executable instructions, where when the computer program or the computer-executable instructions are executed by a processor, the training strategy processing method provided by the embodiment of the present application is implemented.

[0032] The embodiment of the present application has the following beneficial effects:

[0033] Applying the embodiment of the present application, in response to a training strategy customization operation for a first virtual object in a virtual scene corresponding to competitive sports, based on the behavior data of the first virtual object in the virtual scene for competitive interaction, determining a first reference object that matches the first virtual object, obtaining a first training strategy adapted to the first virtual object according to the first reference object, and outputting the first training strategy. Since the first reference object is obtained by matching based on the behavior data of the first virtual object in the virtual scene for competitive interaction, and the first training strategy is obtained according to the first reference object, the first training strategy is adapted to the first virtual object, improving the pertinence of the output training strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is a schematic structural diagram of a training strategy processing system 100 provided by an embodiment of the present application;

[0035] Figure 2 is a schematic structural diagram of an electronic device 500 provided by an embodiment of the present application;

[0036] Figure 3 It is a schematic flowchart of the processing method of the training strategy provided by the embodiments of the present application;

[0037] Figure 4A It is a schematic diagram showing the display of the strategy customization entry provided by the embodiments of the present application Figure 1 ;

[0038] Figure 4B It is a schematic diagram showing the display of the strategy customization entry provided by the embodiments of the present application Figure 2 ;

[0039] Figure 4C It is a schematic diagram showing the display of the strategy customization entry provided by the embodiments of the present application Figure 3 ;

[0040] Figure 5 It is a schematic flowchart of the processing method of the training strategy provided by the embodiments of the present application;

[0041] Figure 6 It is a schematic diagram of the star template provided by the embodiments of the present application;

[0042] Figure 7 It is a matching schematic diagram of the star template provided by the embodiments of the present application;

[0043] Figure 8 It is an analysis schematic diagram of the training strategy provided by the embodiments of the present application;

[0044] Figure 9 It is a schematic diagram of the style analysis report provided by the embodiments of the present application;

[0045] Figure 10 It is a schematic diagram of the training plan provided by the embodiments of the present application. Detailed implementation manners

[0046] 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.

[0047] It can be understood that in the embodiments of the present application, data related to user information and the like 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 relevant laws, regulations and standards.

[0048] 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" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0049] In the following description, the terms "first / second..." are only used to distinguish similar objects and do not represent a specific order for the objects. It is understood that "first / second..." can be interchanged in a specific order or sequence when permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0050] In the embodiments of the present application, the term "module" or "unit" refers to a computer program with a predetermined function or a part of a computer program, which works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as a processing circuit or a memory), or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be a part of an overall module or unit that includes the function of that module or unit.

[0051] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0052] Before further elaborating on the embodiments of the present application, the nouns and terms involved in the embodiments of the present application are described. The nouns and terms involved in the embodiments of the present application are subject to the following explanations.

[0053] 1) Client: An application program running on a terminal for providing various services, such as a video playback client, a game client, etc.

[0054] 2) In response to: Used to represent the conditions or states on which the executed operations depend. When the dependent conditions or states are met, one or more executed operations can be real-time or can have a set delay; without special instructions, there is no limitation on the execution order of the multiple executed operations.

[0055] 3) A virtual scene is a virtual scene displayed (or provided) when an application runs on a terminal. The virtual scene can be a simulation environment of the real world, a semi-simulated and semi-fictional virtual environment, or a purely fictional virtual environment. The virtual scene can be any one of a two-dimensional virtual scene, a 2.5D virtual scene, or a three-dimensional virtual scene. The embodiments of the present application do not limit the dimension of the virtual scene. For example, the virtual scene may include the sky, land, ocean, etc. The land may include environmental elements such as deserts and cities. Users can control virtual objects to move in the virtual scene. For example, the virtual scene can also be used for competitive competitions between at least two virtual objects. For another example, the virtual scene can also be used for competitive competitions between virtual objects in at least two teams.

[0056] Sports competitive games, also known as "activity games", are one of the sports teaching contents and methods. To improve the interest of trainees, a certain sports activity is added with plots or rules, or the result of the activity is used as the basis for judging victory or defeat, so as to improve the enthusiasm of trainees to participate in exercises. The basic elements that make up sports competitive games are physical activities, plots, rules, methods, results, and venues and equipment. Among them, physical activities are indispensable for sports games. Sports competitive games can be classified differently according to teaching requirements and the age characteristics of trainees, such as classification by whether there are plots, the amount of activity, whether to divide into teams or not, and classification by basic human activities such as running, jumping, and throwing. Most popular sports will be included as games, including basketball, tennis, golf, football, table tennis, rugby, boxing, racing, etc. Most sports games involve participating in the game in the form of athletes.

[0057] In sports competitive games, at least two virtual objects conduct a single-round battle mode in the virtual scene. The virtual objects can simulate real sports competitive activities to achieve the purpose of obtaining points in the virtual scene. When the points of the virtual objects in the virtual scene reach certain conditions, the virtual objects win / lose in the virtual scene. Each terminal can control one or more virtual objects in the virtual scene. The competitive mode of sports competitive games can include a single-player competitive mode, a multi-player group competitive mode, or a team large-group competitive mode according to the different corresponding sports competitive activities. The embodiments of the present application do not limit the competitive mode.

[0058] 4) Virtual object: The images of various people and objects that can interact in a virtual scene, or movable objects in the virtual scene. Such movable objects can be virtual characters, virtual animals, anime characters, etc., such as the people, animals, etc. displayed in the virtual scene. A virtual object can be a virtual image in the virtual scene that represents the user. The virtual scene can include multiple virtual objects, and each virtual object has its own shape and volume in the virtual scene, occupying a part of the space in the virtual scene. A virtual object is a game character controlled by the user (or player).

[0059] The embodiments of the present application provide a method, device, electronic device, computer-readable storage medium, and computer program product for processing training strategies, which can improve the pertinence of training strategies. The following describes the exemplary applications of the electronic device provided by the embodiments of the present application. The electronic device provided by the embodiments of the present application can be implemented as various types of user terminals such as laptop computers, tablet computers, desktop computers, set-top boxes, mobile devices (e.g., mobile phones, portable music players, personal digital assistants, dedicated messaging devices, portable game devices), smart phones, smart speakers, smart watches, smart TVs, vehicle terminals, augmented reality (AR) devices, and virtual reality (VR) devices, or can also be implemented as a server. Below, the exemplary applications when the device is implemented as a terminal will be described.

[0060] See Figure 1 , Figure 1 is a schematic diagram of the architecture of the training strategy processing system 100 provided by the embodiments of the present application. To support an exemplary application, the terminals (exemplarily showing terminals 400-1 and 400-2) are 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 the two.

[0061] In some embodiments, an application program (or client) that supports virtual scenes is installed and run on the terminal. The application program can be any one of multiplayer online competitive programs, sports competitive games, and simulation strategy games. In the embodiments of the present application, the application program is taken as an example of a sports competitive game for illustration. When the terminal runs the application program, the user interface of the application program is displayed on the screen of the terminal. The server 200 is the background server corresponding to the application program or client, which can be an independent physical server or a server cluster or distributed system composed of multiple physical servers. The terminal and the server 200 can be directly or indirectly connected through wired or wireless communication methods, which are not limited in the embodiments of the present application.

[0062] In practical applications, the terminal responds to an operation for customizing a training strategy for a first virtual object in a virtual scene corresponding to competitive sports, and sends a training strategy acquisition request to the server 200; based on the training strategy acquisition request, the server 200 extracts features from the behavior data of the first virtual object's competitive interaction in the virtual scene to obtain behavior features characterizing the competitive style of the first virtual object; matches the behavior features with the key features of multiple reference objects, and determines a first reference object that matches the first virtual object among the multiple reference objects based on the matching result; generates a first training strategy adapted to the first virtual object based on the first reference object, obtains the first training strategy adapted to the first virtual object, and returns the first training strategy to the terminal.

[0063] See Figure 2 , Figure 2 is a schematic structural diagram of the electronic device 500 provided by an embodiment of the present application. Taking the terminal in Figure 1 as an example, Figure 2 the electronic device 500 shown includes: at least one processor 510, a memory 550, at least one network interface 520, and a user interface 530. Each component in the electronic device 500 is coupled together through a bus system 540. It can be understood that the bus system 540 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 540 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clear illustration, in Figure 2 all kinds of buses are labeled as the bus system 540.

[0064] The processor 510 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP, Digital Signal Processor), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0065] The memory 550 includes volatile memory or non-volatile memory, and may also include both volatile and non-volatile memory. Among them, the non-volatile memory can be a read-only memory (ROM, Read Only Memory), and the volatile memory can be a random access memory (RAM, Random Access Memory). The memory 550 described in the embodiments of the present application is intended to include any suitable type of memory. Optionally, the memory 550 includes one or more storage devices that are physically far from the processor 510.

[0066] In some embodiments, the memory 550 is capable of storing data to support various operations. Examples of such data include programs, modules, and data structures, or subsets or supersets thereof, which are exemplarily described below.

[0067] The operating system 551 includes system programs for handling 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 handling hardware-based tasks; the network communication module 552 is used to reach other electronic devices via one or more (wired or wireless) network interfaces 520. Exemplary network interfaces 520 include: Bluetooth, Wi-Fi (Wireless Fidelity), and USB (Universal Serial Bus), etc.

[0068] In some embodiments, the processing device of the training strategy provided in the embodiments of the present application can be implemented in software. The processing device of the training strategy provided in the embodiments of the present application can be provided as various software embodiments, including various forms such as application programs, software, software modules, scripts, or code. Figure 2 Shown is the processing device 555 of the training strategy stored in the memory 550, which can be software in the form of programs and plugins, etc., and includes a series of modules, including a reference determination module 5551 and a strategy processing module 5552. These modules are logical, so they can be arbitrarily combined or further split according to the functions to be implemented. The functions of each module will be described below.

[0069] In other embodiments, the device provided in the embodiments of the present application can be implemented in hardware. As an example, the device provided in the embodiments of the present application can be a processor in the form of a hardware decoding processor, which is programmed to execute the processing method of the training strategy provided in 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), digital signal processors (DSPs), programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0070] In some embodiments, a terminal or a server may implement the processing method of the training strategy provided in the embodiments of the present application by running various computer-executable instructions or computer programs. For example, the computer-executable instructions may be microprogram-level commands, machine instructions, or software instructions. The computer program may be a native program or a software module in an operating system; it may be a native application (APPlication, APP), that is, a program that needs to be installed in the operating system to run, such as a game APP or a live broadcast APP; it may also be a small program that can be embedded in any APP, that is, a program that only needs to be downloaded to the browser environment to run. In short, the above computer-executable instructions may be instructions in any form, and the above computer programs may be application programs, modules, or plugins in any form.

[0071] As mentioned above, the processing method of the training strategy provided in the embodiments of the present application may be implemented by various types of electronic devices. For example, it may be Figure 1 individually executed by any one of the terminals and the server 200 in Figure 1 or jointly executed by the terminals and the server 200 in Figure 1 Next, taking the case where the processing method of the training strategy provided in the embodiments of the present application is individually executed by the terminal in Figure 3 as an example for illustration. Refer to Figure 3 which is a schematic flowchart of the processing method of the training strategy provided in the embodiments of the present application, and will be described in conjunction with Figure 3 the steps shown.

[0072] Step 101: In response to a training strategy customization operation for a first virtual object in a virtual scene corresponding to competitive sports, the terminal determines a first reference object that matches the first virtual object based on the behavior data of the first virtual object in the competitive interaction in the virtual scene.

[0073] Among them, the virtual scene corresponding to competitive sports can be the game scenes of competitive sports games such as basketball and football. The training strategy is used to assist the first virtual object in the virtual scene to overcome the competitive disadvantages in competitive interactions, so as to improve the overall competitive level of the first virtual object. Taking the virtual basketball game, a competitive sports game, as an example, the training strategy includes but is not limited to the following strategies: shooting strategy, breakthrough strategy, passing strategy, defense strategy, and rebounding strategy. Among them, the shooting strategy refers to the position, timing, and method of shooting selected by the virtual object (i.e., the virtual star) in competitive interactions (such as games), such as taking a jump shot in an open position after a pick-and-roll or making a quick layup when the defense is lax, so as to improve the shooting percentage and efficiency. The breakthrough strategy refers to the virtual object using techniques such as dribbling and changing directions to break through the opponent's defense and create opportunities for shooting or passing. For example, the breakthrough strategy can be to take a jump shot after a crossover dribble or to directly layup after a change-of-direction move. The passing strategy refers to the virtual star selecting the target, timing, and method of passing in the game to achieve the offensive purpose of the team. For example, the passing strategy can be to pass the ball to a teammate running downcourt in a fast break or to find an open teammate through precise passing in a half-court offense. The defense strategy refers to the star selecting the position, timing, and method of defense in the game to limit the opponent's offense. For example, the defense strategy can be to use full-court pressing to force the opponent into mistakes or to use zone defense to limit the opponent's outside shooting. The rebounding strategy refers to the star's skills and strategies for competing for rebounds in the game to control the game rhythm and ball possession. For example, the rebounding strategy can be to actively jump for offensive rebounds when the opponent shoots or to hold one's position to grab defensive rebounds. These strategies need to be flexibly applied according to the specific situation of the game and the characteristics of the opponent in actual games. Through continuous practice and training, the virtual star can improve its competitive level.

[0074] In some embodiments, the training strategy customization operation is triggered by triggering a strategy customization entry, and the strategy customization entry can be displayed in at least one of the following ways: displaying the strategy customization entry when the competitive interaction of the first virtual object ends; displaying the strategy customization entry in the cultivation training interface of the first virtual object; displaying the strategy customization entry during the competitive interaction of the first virtual object.

[0075] Among them, the strategy customization entry is used to customize the training strategy of the first virtual object in the virtual scene corresponding to competitive sports.

[0076] In practical applications, the processing method of the training strategy provided by the embodiments of the present application can run through multiple stages of the virtual scene corresponding to competitive sports. Taking the virtual scene corresponding to competitive sports as a basketball game as an example,

[0077] the strategy customization entry (such as a button, icon, or link, etc.) can be displayed when the basketball game is in progress, after the basketball game ends, and during the cultivation training of the virtual star. Next, various situations will be described.

[0078] See Figure 4A , Figure 4A which is a schematic diagram showing the display of the strategy customization entry provided by the embodiment of the present application Figure 1 During the game, the first virtual object (corresponding to the current terminal) controlled by the player may encounter various emergencies, such as increased defensive pressure and reduced offensive opportunities. At this time, the strategy customization entry 401 can be displayed on the game interface, allowing the player to obtain the first training strategy generated based on the current interaction progress by triggering the strategy customization entry, so as to instantly adjust the tactics according to the first training strategy and optimize the game performance. For example, when the player discovers a change in the opponent's defensive strategy, they can select the corresponding strategy adjustment by clicking on the strategy customization entry on the screen, such as changing the offensive route or adjusting the defensive position. This instant strategy adjustment can help the player better cope with the game and increase the winning rate.

[0079] See Figure 4B , Figure 4B which is a schematic diagram showing the display of the strategy customization entry provided by the embodiment of the present application Figure 2 After the game, the player can review the performance of the entire game and analyze their competitive advantages (i.e., strengths) and competitive disadvantages (i.e., deficiencies). At this time, providing the strategy customization entry 402 can obtain the first training strategy generated based on the behavior data of the first virtual object throughout the game, so that the player can make more reasonable tactical arrangements for future games according to the first training strategy. For example, if the player finds that they perform poorly in shooting, they can select to strengthen shooting training through the strategy customization entry to improve the shooting percentage. This process of review and summary helps the player continuously improve their competitive level.

[0080] See Figure 4C , Figure 4C which is a schematic diagram showing the display of the strategy customization entry provided by the embodiment of the present application Figure 3 In the training of cultivating virtual star players, the player can train and improve various skills of the first virtual object. At this time, providing the strategy customization entry 403 allows the player to train the competitive skills of the first virtual object specifically according to their competitive style and tactical needs. For example, if the player likes to play fast breaks, they can select to strengthen the speed and passing skills of the star player through the strategy customization entry; if the player likes to play positional warfare, they can choose to strengthen the shooting and rebounding skills of the star player. This personalized training method helps the player build a team that meets their tactical needs.

[0081] Through the above methods, the player can flexibly apply the processing method of the training strategy in different stages and scenarios to improve their game performance and competitive level; at the same time, the design of this strategy customization entry also increases the interactivity and playability of the game, allowing the player to obtain a richer experience in the game.

[0082] In some embodiments, the first reference object matching the first virtual object can be determined based on the behavior data of the first virtual object's competitive interaction in the virtual scene in the following manner: extracting features from the behavior data of the first virtual object's competitive interaction in the virtual scene to obtain behavior features characterizing the competitive style of the first virtual object; matching the behavior features with the key features of multiple reference objects, and determining the first reference object matching the first virtual object from the multiple reference objects based on the matching result.

[0083] In practical applications, in response to a trigger operation for the policy customization entry, the terminal obtains the behavior data of the first virtual object's competitive interaction in the virtual scene. Taking the virtual scene corresponding to competitive sports as a basketball game as an example, the behavior data may include: the number of shots, shooting percentage and position, the number of breakthroughs and success rate, the number of passes, passing accuracy and targets, defensive actions, the number of steals and blocks, the number of rebounds contested and success rate, the movement speed and direction in the game, the decision-making time in the game, and other related behaviors; the behavior features characterizing the competitive style of the first virtual object can be extracted from these behavior data. For example, a player who likes long-range shooting may have a high three-point shooting percentage, while a player who is good at breaking through may have a high layup success rate.

[0084] In some embodiments, the behavior data of the first virtual object's competitive interaction in the virtual scene can be extracted to obtain behavior features characterizing the competitive style of the first virtual object in the following manner: when the behavior data of the first virtual object's competitive interaction in the virtual scene includes the behavior data of multiple competitive actions, performing normalization processing on the behavior data to obtain a behavior data sequence; based on the time dependence of each competitive action in the behavior data sequence, performing encoding processing on the behavior data sequence to obtain a feature vector of the behavior data sequence, where the feature vector indicates at least one of the following action indicators of each competitive action: action type, action frequency, action coherence, action success rate; evaluating the competitive style of the first virtual object based on the feature vector, and determining the behavior features characterizing the competitive style of the first virtual object based on the evaluation result.

[0085] Here, still taking the virtual scenario corresponding to competitive sports as a basketball game as an example, the behavioral data includes at least one of the following: the player behavior sequence (mainly recording the operation process of the first virtual object, such as the coherence of behaviors like passing, dribbling past opponents, dunking, shooting, etc., and these behavior sequences affect the competitive style of the first virtual object), the number of steals (i.e., counting the number of successful steals and the number of times steals are used by the first virtual object in competitive interactions), the number of shooting hits (used to calculate the shooting percentage, where the shooting percentage is equal to the number of goals scored / the number of shots taken), the number of assists (the number of times assisting other virtual objects in the camp where the first virtual object is located), the number of rebounds (including the number of offensive rebounds and defensive rebounds), the number of turnovers (i.e., the number of times being stolen or passing mistakes, etc.), the shooting method (such as three - point shot, mid - range shot, inside - the - paint dunk, layup, or back - to - the - basket play), the holding time (used to calculate the ratio of the time the first virtual object holds the ball to the entire competitive interaction), and the running path (used to measure whether the running of the first virtual object is reasonable. For example, taking the first virtual object as an outside star, the running path is used to measure whether the outside star appears in the positions where it should be during offense and defense).

[0086] After obtaining the behavioral data of the first virtual object, since the data of different competitive actions may have different magnitudes and distributions, the differences can be eliminated by standardizing the behavioral data (such as scaling the data to a fixed range through normalization, for example, between 0 and 1, or converting the data into a distribution with a mean of 0 and a standard deviation of 1 through standardization), so that the data of different competitive actions can be compared and analyzed on the same scale. Through the standardization process, a behavioral data sequence can be obtained, and the data of each competitive action in the behavioral data sequence is comparable.

[0087] Next, based on the time - dependence of each competitive action in the behavioral data sequence, the behavioral data sequence is encoded. Time - dependence means that the occurrence of one competitive action may affect the selection and execution of the next competitive action; the purpose of the encoding process is to capture this time - dependence and convert it into a feature vector. One - hot encoding is performed on the behavioral data sequence to convert each competitive action into a one - hot vector representing the type of the action; if sequence encoding is performed on the behavioral data sequence, considering the order and time interval of the actions, models such as recurrent neural networks or long short - term memory networks are used to encode the sequence data. Through the encoding process, the feature vectors of the behavioral data sequence can be obtained, and these feature vectors indicate indicators such as the action type, action frequency, action coherence, and action success rate of each competitive action.

[0088] Finally, the competitive style of the first virtual object is evaluated based on the feature vectors.

[0089] In some embodiments, the competitive style of the first virtual object can be evaluated based on the feature vector in the following manner: Based on the competitive evaluation rules corresponding to the virtual scene and the feature vector, evaluate the competitive style of the first virtual object to obtain the evaluation result of the competitive style corresponding to the feature vector; correspondingly, the behavioral characteristics representing the competitive style of the first virtual object can be determined based on the evaluation result in the following manner: Determine the standard characteristics corresponding to the competitive style of the feature vector, and use the standard characteristics as the behavioral characteristics representing the competitive style of the first virtual object.

[0090] In practical applications, when evaluating the competitive style of the first virtual object based on the feature vector, a set of competitive evaluation rules is required. These rules define how to derive the competitive style from the feature vector, and these rules are based on the game designer's understanding of the competitive style, or obtained through data analysis and machine learning models. The evaluation process is as follows: Analyze each index in the feature vector to understand the performance of the first virtual object in different competitive actions. For example, compare the data in the feature vector with the competitive evaluation rules to determine whether the performance of the first virtual object in each aspect conforms to the characteristics of a certain competitive style; according to the application results of the rules, generate the evaluation result of the competitive style corresponding to the feature vector. This evaluation result may be a specific competitive style label or an evaluation report containing multiple dimensions. After obtaining the evaluation result of the competitive style, it is necessary to convert these evaluation results into the behavioral characteristics representing the competitive style of the first virtual object. This step involves matching the evaluation results with the predefined standard characteristics.

[0091] The standard characteristics are usually predefined quantitative descriptions of the key aspects of the competitive style. For example, for the shooting percentage, it can be divided into three levels: low, medium, and high. Among them, below 35% is low, 36 - 55% is medium, and above 56% is high; for the number of steals, it can be divided into three levels: few, medium, and many. Among them, less than 1 steal in a single game is few, 2 - 3 steals is medium, and more than 4 steals is many. When matching, match each index in the evaluation result with the standard characteristics to find the standard characteristics that best match the performance of the first virtual object, that is, use the matched standard characteristics as the behavioral characteristics representing the competitive style of the first virtual object. For example, if the evaluation result shows that the shooting percentage of the first virtual object is 60%, then "high shooting percentage" can be used as its behavioral characteristic. If the evaluation result shows that the number of steals of the first virtual object in a single game is 6, then "many steals" can be used as its behavioral characteristic.

[0092] For example, assume that the feature vector of the first virtual object includes indicators such as shooting percentage, breakthrough success rate, and passing accuracy. According to the competitive evaluation rules, it is known that the shooting percentage of the first virtual object is very high, the breakthrough success rate is medium, and the passing accuracy is low. Matching these evaluation results with the standard features gives: shooting percentage: high, breakthrough success rate: medium, passing accuracy: low. Then, the behavioral characteristics representing the competitive style of the first virtual object can be "high shooting percentage, medium breakthrough success rate, and low passing accuracy".

[0093] Through the above method, the competitive style of the first virtual object is evaluated based on the feature vector, and its behavioral characteristics are determined. These behavioral characteristics not only help to understand the competitive style of the first virtual object but also can be used to formulate targeted training plans or training strategies to improve its performance in the virtual scenario.

[0094] In addition, when evaluating the competitive style of the first virtual object based on the feature vector, machine learning or statistical analysis methods can also be used to identify patterns and trends in the feature vector. For example, perform clustering analysis on the feature vector to cluster the feature vector to discover different competitive style groups; or, use a trained classification model to classify the feature vector into predefined competitive style categories. Through the competitive style evaluation, the competitive style of the first virtual object can be determined, and based on the evaluation results, the behavioral characteristics representing its competitive style can be determined. The types of these behavioral characteristics include but are not limited to: action type preference (reflecting which types of competitive actions the first virtual object tends to use), action frequency (the frequency at which the first virtual object performs various competitive actions in the game), action coherence (reflecting the coherence and fluency of the first virtual object when performing a series of competitive actions), and action success rate (reflecting the success rate of the first virtual object when performing various competitive actions).

[0095] Through the above method, by accurately processing and analyzing the behavioral data, the competitive style of the first virtual object is captured, and meaningful behavioral characteristics are provided to provide personalized competitive guidance for the first virtual object to help it achieve better performance in the virtual scenario.

[0096] In practical applications, the reference object can be other players in the game, famous star players in history, or a virtual star player model generated based on game data. Each reference object has its unique competitive style and key features. The so-called key features refer to those features that can significantly affect the competitive style and competitive level of the reference object. These features are usually extracted from a large amount of behavioral data and can reflect the typical behavioral patterns of the reference object in the game. The selection and extraction of key features are based on in-depth analysis of game data and understanding of the competitive style.

[0097] For example, shooting preference reflects the positions (such as outside the three-point line, free throw line, paint area, etc.) where the reference object tends to shoot. If a reference object's number of shots and hit rate outside the three-point line are higher than the average, then "three-point shooting preference" can be used as a key feature of the reference object; for example, if a reference object has a high shooting hit rate (such as more than 50%) and more than 10 shots in a single game, "shooting expert" can be used as a key feature of the reference object. Breakthrough ability reflects the success rate and frequency of the reference object in breaking through the defense. If a reference object often successfully breaks through the defense by changing direction and accelerating, "high breakthrough success rate" can be used as a key feature of the reference object. Passing habits reflect the reference object's preference in passing, such as whether he likes to pass to the inside or outside, or whether he likes to make long passes or short passes. If a reference object often passes high-quality balls to inside stars, "inside pass preference" can be used as a key feature of the reference object. Defensive strategy reflects the reference object's position, movement and interception habits when defending. If a reference object often uses tight defense and can successfully steal the ball, "tight defense strategy" can be used as a key feature of the reference object. Rebounding ability reflects the reference object's sense of position, jumping and positioning skills when competing for rebounds. If a reference object can often occupy a favorable position and successfully grab rebounds in rebounding competitions, "strong rebounding ability" can be used as a key feature of the reference object. Or, if a reference object's number of rebounds in a single game exceeds the first target number (which can be set according to actual needs, such as 7), "rebounding expert" can be used as a key feature of the reference object. Decision-making speed reflects the reference object's decision-making speed in the game, such as how quickly a reference object can make a decision to pass, shoot or break through after receiving the ball. If a reference object can make a decision and execute it quickly after receiving the ball, "quick decision-making ability" can be used as a key feature of the reference object. In addition, if the number of steals of a reference object in a single game exceeds the second target number (which can be set according to actual needs, such as set to 4), then "steal master" can be used as a key feature of the reference object. If the number of mistakes of a reference object in a single game exceeds the third target number (which can be set according to actual needs, such as set to 5), then "mistake master" can be used as a key feature of the reference object. If a reference object has a single-game mid-range shot hit rate of up to 50% in a single game and makes 4 mid-range shots, then "mid-range shot handsome boy" can be used as a key feature of the reference object. Of course, the key features of the reference object can also be determined from other aspects, and the embodiment of the present application does not limit the method for determining the key features. The key features of the reference object can help us understand the competitive style of the reference object and provide personalized training strategy guidance for the first virtual object. By analyzing and applying these features, the performance and competitiveness of the first virtual object in the game can be improved.

[0098] In some embodiments, the behavioral characteristics and the key characteristics of multiple reference objects can be matched in the following manner: calculate the similarity between the behavioral characteristics and the key characteristics of each reference object, and use the similarity between the behavioral characteristics and the key characteristics of each reference object as the matching result; correspondingly, the first reference object that matches the first virtual object can be determined from multiple reference objects based on the matching result in the following manner: screen out the reference object corresponding to the maximum similarity among multiple reference objects, and use the screened reference object as the first reference object that matches the first virtual object.

[0099] In practical applications, when matching the behavioral characteristics of the first virtual object and the key characteristics of multiple reference objects, machine learning algorithms (such as clustering, classification, or regression) can be used to match the behavioral characteristics of the first virtual object and the key characteristics of the reference objects, such as calculating the similarity or distance between the behavioral characteristics of the first virtual object and the key characteristics of the reference objects to determine the matching degree between the first virtual object and the reference objects, and determining the first reference object that is most similar to the first virtual object according to the matching degree between the first virtual object and each reference object, such as using the reference object with the maximum similarity as the first reference object that matches (i.e., is most similar to) the first virtual object.

[0100] After determining the first reference object that matches the first virtual object, a first training strategy adapted to the first virtual object can be generated based on the competitive style and training strategy of the first reference object. This may include: suggesting that the first virtual object increase shooting practice at specific positions according to the shooting habits of the first reference object, providing guidance and suggestions for breakthrough training to the first virtual object according to the breakthrough skills of the first reference object, providing improvement methods for defensive skills to the first virtual object according to the defensive strategy of the first reference object, and providing suggestions for rebounding training to the first virtual object according to the rebounding habits of the first reference object. In addition, the first training strategy is also customized according to the specific behavioral characteristics and game requirements of the first virtual object to help players improve their game performance and competitive level.

[0101] Through the above method, by accurately extracting and understanding behavioral characteristics, and effectively matching and applying the strategies of reference objects, personalized training strategy guidance can be provided to players to help them achieve better results in virtual games.

[0102] Step 102: Obtain a first training strategy adapted to the first virtual object according to the first reference object, and output the first training strategy.

[0103] In some embodiments, the first training strategy adapted to the first virtual object can be obtained based on the first reference object in the following manner: obtain the reference training strategy of the first reference object and use the reference training strategy as the first training strategy adapted to the first virtual object.

[0104] In practical applications, after determining the first reference object that matches the first virtual object, the reference training strategy corresponding to the first reference object can be obtained as the first training strategy configured for the first virtual object. For example, if the first reference object that matches the first virtual object is star player A, and his corresponding competitive style is a three-point finisher, his advantage is the ability to have a stable scoring output; his weakness is that his footwork is relatively slow and he needs to have his own rhythm to score relatively highly. The way to improve is how to strengthen the offense. Then, the reference training strategy of star player A is used as the first training strategy. For example, the training strategy of a three-point finisher mainly focuses on three-point shooting offense and includes four aspects: shooting timing, getting rid of defenders, using screens, and the exclusive training of star player A. Among them, the training strategy for shooting timing mainly focuses on open shots, looking for open shots, etc. The training strategy for getting rid of defenders mainly uses one's own offensive dribbling. The training strategy for using screens mainly uses the running and cooperation of teammates. The training strategy of star player A includes suggestions on star player A's own dribbling + fakes + shooting.

[0105] Through the above method, players can obtain a training strategy adapted to themselves, so as to use the training strategy specifically in competitive interactions. This facilitates players to better familiarize themselves with the game, the star players, and the playing styles, and enables them to understand their own styles, so as to better play to their own and the star players' advantages to win for the team.

[0106] In some embodiments, the first training strategy adapted to the first virtual object can be obtained based on the first reference object in the following manner: determine the competitive disadvantages of the first virtual object based on the behavioral characteristics of the first virtual object, and generate a strengthening training strategy for overcoming the competitive disadvantages; determine the reference training strategy of the first reference object, and fuse the reference training strategy and the strengthening training strategy to obtain the first training strategy adapted to the first virtual object.

[0107] In practical applications, when generating the first training strategy for the first virtual object, first, conduct an in-depth analysis of the behavioral characteristics of the first virtual object to identify its performance in different competitive aspects. For example, by comparing its behavioral characteristics with the standard characteristics, determine in which aspects it has disadvantages. For example, if the passing accuracy of the first virtual object is low, then passing is one of its competitive disadvantages. Then, based on the identified competitive disadvantages, generate targeted strengthening training strategies. For example, for the disadvantage of low passing accuracy, a strategy to strengthen passing training can be generated, including improving passing skills and enhancing passing awareness. Then, determine the reference training strategy of the first reference object, analyze the training strategy of the first reference object, and extract its advantages and characteristics in various aspects. Finally, fuse the reference training strategy and the strengthening training strategy to obtain the first training strategy adapted to the first virtual object. Among them, the fusion process needs to consider the existing competitive style and advantages of the first virtual object, as well as the strategy characteristics of the reference object, to ensure that the fused training strategy can fully exert the potential of the first virtual object while making up for its competitive disadvantages.

[0108] Through the above method, by analyzing the competitive disadvantages of the first virtual object and generating personalized strengthening training strategies, the weak links of the first virtual object can be targeted for improvement; by fusing the reference training strategy of the reference object, the first virtual object can draw on the strategy characteristics of excellent star players, which can not only improve the overall competitive level of the virtual object in the game, but also provide a personalized gaming experience and enhance the player's participation and satisfaction.

[0109] In some embodiments, after determining the first reference object that matches the first virtual object, a competitive style analysis interface of the first virtual object can be displayed based on the first reference object; in the competitive style analysis interface, display the competitive style of the first virtual object, the first reference object that matches the first virtual object, the competitive advantages and disadvantages of the first reference object, and training tips for overcoming the competitive disadvantages.

[0110] In actual applications, after determining the first reference object that matches the first virtual object, the competitive style of the first virtual object is consistent with the competitive style of the first reference object, and the competitive style analysis interface of the first virtual object can be displayed based on the competitive style of the first reference object. The player can easily view and understand the competitive style-related information of the first virtual object in the competitive style analysis interface, such as displaying the competitive style label or description of the first virtual object in the competitive style display area of ​​the competitive style analysis interface, such as "offensive", "defensive", "all-around", "three-point terminator", etc.; displaying the image and basic information of the first reference object that matches the first virtual object in the reference object display area, and introducing its performance and characteristics in the game; displaying the competitive advantages and disadvantages of the first reference object in the competitive advantages and disadvantages display area; and displaying training tips for overcoming competitive disadvantages in the training strategy display area, that is, providing specific training strategy suggestions, including training methods, tactical adjustments, etc., to help the first virtual object overcome its competitive disadvantages.

[0111] In the above manner, the player can clearly understand the competitive style, competitive advantages, competitive disadvantages and improvement suggestions of the first virtual object through the competitive style analysis interface, and perform corresponding competitive actions based on the improvement suggestions to promote its growth and improvement in the game.

[0112] In some embodiments, the first training strategy can be output in the following manner: displaying a training prompt interface for the first virtual object; in response to the first training strategy including sub-training strategies for multiple training items, displaying a training entry for each training item in the training prompt interface; in response to a selection operation, outputting the sub-training strategy for the training item corresponding to the selected training entry.

[0113] In actual applications, the training prompt interface is intended to show players the training suggestions and entrances for the first virtual object. The training prompt area of ​​the training prompt interface can display training suggestions and guidance related to the first training strategy; if the first training strategy includes multiple training items, the entrance button or link of each training item can be displayed separately in the training item entrance area, and each training entrance should clearly indicate the content and purpose of the training item, such as "shooting training", "breakthrough training", "passing training", etc. When a player clicks on the training entrance of a training item, the system should output the sub-training strategy of the training item corresponding to the selected training entrance. The sub-training strategy should include specific training steps, skill guidance, training goals, etc., to help players conduct effective training.

[0114] For example, if the first reference object matching the first virtual object is star player A and his corresponding competitive style is three-point finisher, the reference training strategy of star player A is used as the first training strategy. For example, the training strategy of a three-point finisher mainly focuses on three-point shooting offense. The following training entrances can be displayed: the training entrance for shooting timing, the training entrance for getting rid of defense, the training entrance for using pick-and-roll, and the exclusive training entrance for star player A. When the player triggers the training entrance for shooting timing, the terminal responds to this trigger operation and outputs corresponding sub-training strategies, such as shooting timing training mainly carried out around open shots, looking for open shots, etc. When the player triggers the training entrance for getting rid of defense, the terminal responds to this trigger operation and outputs corresponding sub-training strategies, such as the training strategy for getting rid of defense by using one's own offensive dribbling, and so on.

[0115] Through the above method, through the training prompt interface, players can obtain personalized training guidance for the competitive disadvantages of the first virtual object, which helps to improve their competitive level; the setting of training item entrances enables players to easily access different training items and saves the time for finding training content; by selecting an operation to output specific sub-training strategies, players can interactively learn and master different training skills and improve the learning effect; personalized training guidance and interactive learning experience can enhance players' immersion in the game and make the game more interesting and challenging.

[0116] In some embodiments, the training entrances of each training item can be displayed in the training prompt interface in the following way: determine the influencing factors for affecting the training priority of the training item, and the influencing factors include at least one of the following: required duration, difficulty, type, and synergistic effect on other training items; predict the training priority of each training item based on the influencing factors, and display the training entrances of each training item in the training prompt interface in a manner corresponding to the training priority.

[0117] Here, when there are at least two training items, the training priorities corresponding to different training items can be different. The training entrances of each training item can be displayed in corresponding display styles according to the training priorities corresponding to the training items, that is, for training items with different training priorities, different display styles (such as different display orders, different display colors, different display brightnesses, different display fonts, etc.) are used to display the training entrances of each training item, or the training entrances of training items with higher training priorities are highlighted. For example, there are 3 selectable training items, namely training item 1, training item 2, and training item 3. The training priorities corresponding to these three training items decrease in turn. The training entrance of training item 1 can be displayed in red font, the training entrance of training item 2 can be displayed in yellow font, and the training entrance of training item 3 can be displayed in gray font, which is convenient for users to select the training entrances of the required training items and is conducive to the pertinence of training.

[0118] In actual implementation, the training priority of a training item can be determined through the following process: Obtain the influencing factors that affect the training priority of the training item. Based on the influencing factors, call a neural network model to perform a training priority prediction process to obtain the training priority of the training item. Taking the virtual scenario corresponding to competitive sports as a basketball game as an example, determine influencing factors such as the duration required to complete the training strategy corresponding to a training item, the difficulty of the training item (such as low, medium, high, etc.), the type of the training item (such as shooting, passing, defense, rebounding, etc.), and the synergy effect on other training items. Among them, the synergy effect is illustrated by the following example: Suppose that after training item A is well-trained, it will be very helpful for the training of training item B. Then the influencing factor of training item A on training item B can be summarized as "synergy effect". This influencing factor reflects how the mastery of training item A promotes or affects the learning and progress of training item B. For example, training item A may involve the training of basic skills, and these basic skills are also important in training item B. Therefore, the proficiency of training item A will directly improve the training results of training item B. This relationship can be used to determine the priority of training items to ensure that players first master those skills that can provide support and help for subsequent training.

[0119] After obtaining the influencing factors that affect the training priority of the training item, inputting the influencing factors into the trained neural network model for prediction processing to obtain the training priority of the training item can make the prediction more accurate. Before applying the neural network model, it is necessary to train the initial neural network model and then put the trained neural network model into application to realize combining the influencing factors and predicting the training priority of the training item through artificial intelligence technology. Among them, the neural network model is trained through training item samples, the influencing factors corresponding to the training item samples, and the training priorities annotated for the training item samples. For example, based on the training item samples, the synergy effect associated with the training item samples, and the training priorities annotated for the training item samples, call the initial neural network model to perform prediction processing to obtain the predicted training priority. After determining the value of the loss function of the neural network model through the predicted training priority and the annotated training priority, it can be judged whether the value of the loss function exceeds the preset threshold. When the value of the loss function exceeds the preset threshold, determine the error signal of the neural network model based on the loss function, backpropagate the error signal in the neural network model, and update the model parameters of each layer during the propagation process.

[0120] Among them, the embodiments of the present application do not limit the model structure of the neural network model. For example, the neural network model can be a convolutional neural network, a deep neural network, etc.; nor are they limited to the form of the loss function. For example, it can be a cross-entropy loss function, an L2 loss function, etc. Here, the backpropagation is described. The training sample data is input into the input layer of the neural network model, passes through the hidden layer, and finally reaches the output layer and outputs the result. This is the forward propagation process of the neural network model. Since there is an error between the output result of the neural network model and the actual result, the error between the output result and the actual value is calculated, and this error is propagated backward from the output layer to the hidden layer until it reaches the input layer. During the backpropagation process, the values of the model parameters are adjusted according to the error, that is, a loss function is constructed based on the error between the output result and the actual value, and the partial derivatives of the loss function with respect to the model parameters are calculated layer by layer to generate the gradients of the loss function with respect to the model parameters of each layer. Since the direction of the gradient indicates the direction in which the error expands, the gradient of the model parameters is taken as the opposite, and summed with the original parameters of each layer of the model. The obtained summation result is used as the updated model parameters of each layer, so as to reduce the error caused by the model parameters, and continuously iterate the above process until convergence.

[0121] In some embodiments, the first training strategy can be output in the following manner: in response to satisfying the strategy prompt condition, the first training strategy is output in the form of strategy prompt information; wherein, the strategy prompt information includes the target competitive action recommended to be executed based on the first training strategy and the timing of executing the target competitive action.

[0122] In practical applications, the strategy prompt condition can be preset. The strategy prompt condition can be triggered based on specific situations in the game or the behavioral characteristics of the first virtual object. During the competitive interaction of the first virtual object, when the strategy prompt condition is satisfied, the system automatically generates strategy prompt information. These information should be concise and clear, and directly guide the player to execute specific competitive actions (i.e., target competitive actions), that is, the strategy prompt information should include the target competitive action recommended to be executed based on the first training strategy and the timing of executing this action. For example, the prompt information can be: "There is a current defensive gap, it is recommended to shoot immediately!", where the target competitive action is "shoot", and the execution timing is "immediately, before the defense recovers".

[0123] It should be noted that the strategy prompt information can be displayed to the player through a specific area of the game interface (such as the HUD heads-up display) or a pop-up window, ensuring that the strategy prompt information does not interfere with the player's line of sight, while being easy to identify and understand.

[0124] In this way, the strategy prompt information can timely guide the player to execute the most favorable competitive actions, improving the performance and competitiveness of the first virtual object in the game.

[0125] In some embodiments, the satisfaction of the policy prompt condition can be determined in the following manner: when a policy prompt entry is displayed in the interface of the virtual scenario, in response to a trigger operation on the policy prompt entry, it is determined that the policy prompt condition is satisfied; or, in response to the interaction progress of the first virtual object in the competitive interaction reaching the target progress, it is determined that the policy prompt condition is satisfied, and the target progress includes at least one of the following: there is an association between the competitive action currently executed by the first virtual object and the target competitive action recommended by the first training policy, and the competitive disadvantage targeted by the first training policy appears in the first virtual object.

[0126] In practical applications, the player can actively request to obtain the first training policy. In this case, a policy prompt entry (such as a button, icon, or link) can be displayed in the interface of the virtual scenario. The player requests to obtain the first training policy by clicking or selecting this policy prompt entry. That is, when the player actively triggers the policy prompt entry, the terminal detects this trigger operation, determines that the policy prompt condition is satisfied, and then outputs the policy prompt information of the first training policy.

[0127] Taking the virtual scenario corresponding to competitive sports as a football game as an example, assume that the first virtual object is a forward in a football game, and its training policy emphasizes shooting when the opponent's defense is weak. When the player triggers the policy prompt entry in the game interface, the terminal responds to this trigger operation and outputs the policy prompt information: "There is a current defensive gap. It is recommended to shoot immediately!", where the target competitive action is "shoot", and the execution timing is "immediately, before the defense recovers".

[0128] In addition, the first training policy can also be automatically output by the terminal based on the interaction progress of the first virtual object. For example, the terminal real-time detects the interaction progress of the first virtual object in the competitive interaction. When the interaction progress reaches the preset target progress, the system automatically determines that the policy prompt condition is satisfied, where the interaction progress can be that there is an association between the competitive action currently executed by the first virtual object and the target competitive action recommended by the first training policy. Taking the virtual scenario corresponding to competitive sports as a football game as an example, assume that the first virtual object is a forward, and its training policy recommends shooting when the opponent's defense is weak. Currently, the first virtual object is dribbling through and approaching the opponent's goal, and the number of opponent's defensive star players is small. In this scenario, the terminal detects that the dribbling through action of the first virtual object is associated with the shooting target action in the training policy, and considers that the interaction progress of the first virtual object in the competitive interaction reaches the target progress, that is, the policy prompt condition is satisfied. At this time, the terminal automatically directly outputs the policy prompt information: "The defense is empty. It is recommended to shoot immediately!", where the target competitive action is "shoot", and the execution timing is "immediately, before the defense recovers".

[0129] Taking the virtual scenario corresponding to competitive sports as an example of a basketball game, assume that the first virtual object is a star player in a basketball game, and its training strategy emphasizes shooting when the defense is weak. Then, when the first virtual object has the ball, is outside the three-point line, the opponent's defense of the star player is temporarily out of position, and a defensive gap appears, and the terminal detects the defensive gap and the first virtual object is outside the three-point line, it is considered that the interaction progress of the first virtual object in the competitive interaction reaches the target progress, that is, the strategy prompt condition is satisfied. At this time, the terminal automatically directly outputs the strategy prompt information: "There is a current defensive gap, it is recommended to shoot immediately!", where the target competitive action is "shooting", and the execution timing is "immediately, before the defense recovers".

[0130] The interaction progress can also be that the competitive disadvantage targeted by the first training strategy appears in the first virtual object. Taking the virtual scenario corresponding to competitive sports as an example of a basketball game, assume that the first virtual object is a guard, and the competitive disadvantage targeted by its training strategy is a low breakthrough success rate. In the current game, if the first virtual object continuously tries to break through but fails, and the breakthrough success rate drops significantly, that is, the competitive disadvantage appears, it is considered that the interaction progress of the first virtual object in the competitive interaction reaches the target progress, that is, the strategy prompt condition is satisfied. At this time, the terminal automatically outputs the strategy prompt information: "The breakthrough is blocked, it is recommended to try passing or shooting.", where the target competitive action is "passing or shooting", and the execution timing is "immediately", to prompt the player to change the offensive method to overcome the breakthrough disadvantage.

[0131] In the above way, players can actively request to obtain strategy prompts, actively learn and master game strategies; the terminal can automatically output strategy prompts according to the interaction progress, providing real-time targeted guidance and suggestions; real-time strategy prompts and active learning opportunities can enhance the player's game immersion and make them more involved in the game.

[0132] In some embodiments, after outputting the first training strategy, the first training strategy can be updated in the following way: in the case where the first virtual object executes the first training strategy multiple times to overcome the competitive disadvantage, determine the improvement degree of each execution of the first training strategy for the competitive disadvantage; fuse the improvement degrees of each execution of the first training strategy for the competitive disadvantage to obtain a comprehensive improvement degree; in response to the comprehensive improvement degree being lower than the degree threshold, replace the first training strategy with a second training strategy, where the second training strategy is generated based on the skill style of the second reference object, and the similarity between the behavioral characteristics of the first virtual object and the key characteristics of the second reference object is the same as the similarity between the behavioral characteristics of the first virtual object and the key characteristics of the first reference object.

[0133] In practical applications, the terminal continuously detects the performance of the first virtual object in the competitive interaction, especially its execution when the interaction progress reaches the target progress. Each time the first virtual object reaches the target progress and executes the first training strategy, the degree of improvement in its competitive disadvantage is evaluated, which can be achieved by comparing the performance data before and after executing the first training strategy, such as the increase in breakthrough success rate, the improvement in shooting percentage, etc. Then, the degrees of improvement obtained from multiple executions of the first training strategy are fused (e.g., by taking the average, weighted summation, or other statistical methods) to calculate a comprehensive degree of improvement. Finally, the comprehensive degree of improvement is compared with a preset degree threshold. If the comprehensive degree of improvement is lower than the degree threshold, it indicates that the first training strategy is not effective in improving the competitive disadvantage. In this case, the terminal decides to replace the first training strategy with the second training strategy, where the second training strategy is generated based on the skill style of the second reference object, and the selection of the second reference object is based on the similarity between its key features and the behavioral features of the first virtual object. Specifically, the similarity between the behavioral features of the first virtual object and the key features of the second reference object should be the same as the similarity between the behavioral features of the first virtual object and the key features of the first reference object, which means that the second reference object has similarities with the first virtual object in certain skills or styles, but may be different in other aspects, thus providing different strategy options. The terminal provides the second training strategy to the player as a new option to improve the competitive disadvantage, and the player can train and practice according to the prompts of the new second training strategy in the hope of seeing better results.

[0134] Taking a football game in the virtual scenario corresponding to competitive sports as an example, assume that the first virtual object controlled by the player is a striker, and his competitive disadvantage is a low shooting accuracy. The first training strategy generated by the game system for him is: "When receiving a pass in front of the opponent's penalty area, it is recommended to shoot immediately." When the terminal detects the competitive interaction of the first virtual object, whenever the first virtual object receives a pass in front of the penalty area (the interaction progress reaches the target progress), the terminal will record whether he shoots according to the strategy and the result of the shot (whether it is a goal or a miss). The terminal evaluates the contribution degree of each shot of the first virtual object to improving the shooting accuracy. For example, if the first virtual object shoots according to the first training strategy and scores a goal, it is considered that the improvement degree corresponding to the first training strategy is high; if the first virtual object shoots according to the first training strategy but misses, it is considered that the improvement degree corresponding to the first training strategy is low or negative. The terminal fuses the improvement degrees of the first virtual object in multiple shots and calculates a comprehensive improvement degree. Assume that after five games or the first virtual object shoots 5 times according to the first training strategy, the comprehensive improvement degree is a slight improvement. The terminal compares the comprehensive improvement degree (slight improvement) with a preset degree threshold (for example, significant improvement). If the comprehensive improvement degree is lower than the degree threshold, the terminal selects a second reference object. The second reference object has similarities with the first reference object in shooting skills, but may have different styles in other aspects such as passing or dribbling. Based on the skill style of the second reference object, a second training strategy is generated: "When receiving a pass in front of the opponent's penalty area, it is recommended to first observe the position of the defensive star and then choose the best shooting angle", and a prompt can be sent to the player: "The current strategy has limited effect on improving the shooting accuracy. It is recommended to try a new strategy." The player can choose to accept the new strategy (i.e., the second training strategy) and guide the first virtual object to execute according to the new strategy in the next game.

[0135] In the above way, when the player's multiple executions or practices according to the training strategy suggestions do not significantly improve the competitive disadvantage of the first virtual object, a prompt to change the strategy is given in a timely manner. By detecting, evaluating, and dynamically adjusting, a second training strategy more suitable for the first virtual object is generated, and it is recommended that the player change the first training strategy to the second training strategy to help the first virtual object overcome the competitive disadvantage. This dynamic adjustment of the training strategy according to the performance of the first virtual object ensures the timeliness and effectiveness of the training strategy, and helps to improve the player's competitive performance and the player's game experience.

[0136] In some embodiments, after outputting the first training strategy, the first training strategy can be updated in the following manner: in response to an operation of using the first training strategy, control the first virtual object to execute the first training strategy in the competitive interaction; during the process of the first virtual object executing the first training strategy, detect the competitive style of the first virtual object based on new behavior data; in the case where there is a difference between the current competitive style and the original competitive style, adjust the first training strategy so that the adjusted first training strategy is adapted to the current competitive style.

[0137] In practical applications, after the terminal determines the first training strategy suitable for the first virtual object, in the competitive interaction, the player controls the first virtual object to execute corresponding competitive actions according to the first training strategy. For example, if the first training strategy is "shoot when the opponent's defense is weak", the player will guide the first virtual object to take a shooting action when the right opportunity appears in the game. During the process of the first virtual object executing the first training strategy, the terminal collects the behavior data of the first virtual object in real time, and analyzes and detects whether the competitive style of the first virtual object has changed based on these behavior data. If the terminal detects a significant difference between the current competitive style and the original competitive style of the first virtual object, it will automatically adjust the first training strategy to ensure that the strategy can better adapt to the current competitive style of the virtual object. This adjustment may be a fine-tuning of some parameters of the first training strategy, or a complete change in the direction of the first training strategy. The adjusted first training strategy will be output to the player, and the player can continue to guide the competitive behavior of the first virtual object according to the new first training strategy. This dynamic strategy adjustment helps the virtual object maintain the best performance in the competitive interaction and improves the player's gaming experience at the same time.

[0138] Taking a basketball game in a virtual scenario corresponding to competitive sports as an example, the first virtual object controlled by the player is a guard, and his competitive disadvantage is a low breakthrough success rate. The game system generates a first training strategy for him: "When there is a gap in the opponent's defense line, it is recommended to immediately make a breakthrough.", and outputs the first training strategy to the player, suggesting that the first virtual object immediately make a breakthrough when there is a gap in the opponent's defense line. During the game, the player controls the first virtual object to immediately make a breakthrough when there is a gap in the opponent's defense line according to the first training strategy. The terminal collects the behavioral data of the first virtual object during the execution of the strategy. If it is found that after multiple attempts to break through, his breakthrough success rate has increased, but the number of breakthrough attempts has decreased, indicating that his competitive style has become more cautious, which means that the competitive style of the first virtual object has changed, from the original aggressive breakthrough to a more cautious breakthrough style. Therefore, the terminal adjusts the first training strategy to: "When there is a gap in the opponent's defense line, it is recommended to first observe the movement of the defensive star and then select the best breakthrough timing.", prompts the player with the adjusted first training strategy, and suggests that the player guide the competitive behavior of the first virtual object according to the new first training strategy. The player can control the first virtual object to take actions more suitable for the current competitive style in the next game according to the adjusted first training strategy, such as first observing the movement of the defensive star and then selecting the best breakthrough timing.

[0139] In the above way, the terminal detects the change in the competitive style of the first virtual object by detecting the behavioral data of the first virtual object during the execution of the first training strategy, and adjusts the training strategy accordingly to ensure that the strategy is adapted to the current competitive style. This dynamic adjustment and personalized guidance help to improve the competitive performance of the first virtual object and the game experience of the player.

[0140] In some embodiments, after outputting the first training strategy adapted to the first virtual object, the first training strategy can be updated in the following way: in response to a feedback operation, obtain feedback information for the first training strategy; update and adjust the first training strategy based on the feedback information.

[0141] In practical applications, the terminal outputs a first training strategy specifically designed for the first virtual object based on its behavioral characteristics and competitive needs. The first training strategy aims to help the first virtual object overcome its competitive disadvantages or leverage its advantages. The terminal can provide a feedback channel so that players can subjectively evaluate the effectiveness of the strategy after executing the first training strategy. The feedback can be positive (such as "the strategy is very effective") or negative (such as "the strategy has poor effects"), or it can be specific suggestions (such as "hoping that the strategy can consider more defensive factors"). The terminal collects the feedback information from players and incorporates it into the model update process. This feedback can help the system understand the actual effectiveness of the strategy, as well as the specific needs and preferences of players. The terminal uses the feedback information to update and adjust the first training strategy to improve its practicality and personalization. This may include adjusting the parameters of the strategy, changing the direction of the strategy, or completely redesigning the strategy. By continuously iterating the key characteristics of the reference object (i.e., the star data and star characteristics in the star model library) to reflect player feedback and strategy adjustments, it helps to make the star model library more comprehensive and accurate, better meeting the individual needs of players. The updated first training strategy is output to the players, and the players can continue to guide the competitive behavior of the first virtual object according to the new first training strategy. This personalized guidance based on feedback helps to improve the competitive performance of the virtual object and enhance the gaming experience of players.

[0142] Taking the virtual scenario corresponding to competitive sports, such as a football game, as an example. Suppose the first virtual object controlled by the player is a striker, and his competitive disadvantage is a low shooting accuracy. The game system generates a first training strategy for him: "When receiving a pass in front of the opponent's penalty area, it is recommended to shoot immediately.", and outputs the first training strategy to the player, suggesting that the striker shoot immediately when receiving a pass in the penalty area. After the player controls the first virtual object to execute the first training strategy, the player submits feedback information through the in-game feedback system. For example, the player may feedback: "The strategy has average effects, and the striker often misses the goal when shooting." The game system collects the feedback information from the player and analyzes the reasons. For example, it may be that the first virtual object lacks sufficient preparation when shooting, or interference from defensive stars causes the shot to miss. The system adjusts the first training strategy according to the feedback information to: "When receiving a pass in front of the opponent's penalty area, it is recommended to adjust the pace first and then choose the best shooting opportunity.", and updates the striker data in the star model library to reflect the improvement in his adjustment ability and shooting opportunity selection when shooting. The system prompts the player with the adjusted first training strategy and suggests that the player guide the competitive behavior of the first virtual object according to the new first training strategy. The player can control the first virtual object to take more suitable shooting actions according to his personal characteristics in the next game according to the adjusted first training strategy, such as adjusting the pace first and then choosing the best shooting opportunity.

[0143] In the above - mentioned manner, by providing a feedback channel to collect players' subjective evaluations and incorporating them into the model update process, the star data and star characteristics in the star model library are continuously iterated to provide more personalized and effective training strategy guidance. This feedback - based dynamic adjustment mechanism helps to improve the competitive performance of virtual objects and the gaming experience of players.

[0144] In some embodiments, after outputting the first training strategy adapted to the first virtual object, the first virtual object can be motivated to actively participate in the training of the first training strategy in the following way: creating a training challenge task for the first training strategy and publishing the training challenge task to the terminal of the second virtual object, where the second virtual object is matched with the first target object; in response to the second virtual object accepting the training challenge task, comparing the training results of the first virtual object for the first training strategy with the training results of the second virtual object for the first training strategy.

[0145] In practical applications, the terminal outputs a first training strategy specifically designed for the first virtual object according to the behavior characteristics and competitive needs of the first virtual object. The first training strategy aims to help the first virtual object overcome its competitive disadvantages or give full play to its advantages. The terminal can create a training challenge task for the first training strategy, and the task content may include specific competitive action practice, skill improvement goals, etc., and publish the training challenge task to the terminal of the second virtual object that matches the first virtual object. The matching can be based on factors such as competitive level, gaming style, geographical location, etc. When the second virtual object accepts the training challenge task, the system records this acceptance behavior and prepares for subsequent comparison of training results. The first virtual object and the second virtual object execute the first training strategy in their respective trainings and generate training result data. The system collects and compares the training results of both to evaluate their progress and performance in executing the first training strategy. The system allows players to share their training achievements and progress, and can create a leaderboard. Players can participate in these activities to showcase their competitive levels and compete with other players.

[0146] Taking the virtual scenario corresponding to competitive sports as an example of a basketball game, assume that the first virtual object controlled by the player is a guard, and his competitive disadvantage is a low breakthrough success rate. The game system generates a first training strategy for him: "When there is a gap in the opponent's defense line, it is recommended to break through immediately." The system outputs the first training strategy to the player, suggesting that the guard break through immediately when there is a gap in the opponent's defense line. The system creates a training challenge task: "In the next five games, try to break through at least ten times and record the success rate." and publishes this task to the terminal of the second virtual object that matches the first virtual object. The second virtual object may be a guard in the same position, with a competitive level similar to that of the first virtual object. The second virtual object accepts the challenge, the system records the acceptance behavior, and prepares to compare the training results. The first virtual object and the second virtual object execute the breakthrough strategy in their respective five games and record the success rate. The system collects the breakthrough times and success rate data of both, conducts comparison analysis and display. The display of the comparison results and the leaderboard can stimulate the player's sense of achievement and competitive desire, and further encourage them to participate in training. Through social interaction and competition, the system motivates players to actively participate in training challenge tasks and improve their competitive level.

[0147] In the above manner, by creating a training challenge task, publishing it to the matching virtual object, comparing the training results, and providing social interaction and mechanisms, the first virtual object is motivated to actively participate in the training of the first training strategy. This incentive mechanism helps to improve the competitive performance of the virtual object and the player's gaming experience.

[0148] Next, an exemplary application of the embodiments of the present application in an actual application scenario will be described. Taking the virtual scenario corresponding to competitive sports as an example of a basketball game, the processing method of the training strategy provided by the embodiments of the present application will be further described. Refer to Figure 5 , Figure 5 which is a schematic flowchart of the processing method of the training strategy provided by the embodiments of the present application. The method includes the following key steps:

[0149] Step 201: Data collection and preprocessing.

[0150] First, the behavior data of the first virtual object (the player corresponding to the current terminal) during the competitive interaction in the basketball game is collected in real time. These data reflect the performance of the player. The behavior data includes, but is not limited to: the player behavior sequence (mainly recording the operation process of the first virtual object, such as the coherence of behaviors like passing, dribbling past opponents, dunking, shooting, etc. These behavior sequences affect the competitive style of the first virtual object), the number of steals (i.e., counting the number of successful steals and the number of times steals are used by the first virtual object during the competitive interaction), the number of shooting hits (used to calculate the shooting percentage, where the shooting percentage is equal to the number of goals scored / the number of shots taken), the number of assists (the number of times assisting other virtual objects in the camp where the first virtual object is located), the number of rebounds (including the number of offensive rebounds and defensive rebounds), the number of turnovers (i.e., the number of times being stolen or passing mistakes, etc.), the shooting method (such as three-point shot, mid-range shot, inside dunk, layup, or back-to-back play), the holding time of the ball (used to calculate the ratio of the duration that the first virtual object holds the ball in the entire competitive interaction), and the running path (used to measure whether the running of the first virtual object is reasonable. For example, taking the first virtual object as an outside star, the running path is used to measure whether the outside star appears at the positions where it should be during offense and defense).

[0151] After obtaining the behavior data of the first virtual object, since the data of different competitive actions may have different magnitudes and distributions, the differences can be eliminated by cleaning and standardizing the behavior data (such as scaling the data to a fixed range through normalization, for example, between 0 and 1, or converting the data to a distribution with a mean of 0 and a standard deviation of 1 through standardization), so that the data of different competitive actions can be compared and analyzed on the same scale. Through the standardization process, a behavior data sequence can be obtained, and the data of each competitive action in the behavior data sequence is comparable.

[0152] Step 202: Feature extraction.

[0153] This step is mainly used to extract the feature vectors that can effectively represent the competitive style and competitive level of the first virtual object. Specifically, first, based on the time dependence of each competitive action in the behavior data sequence, the behavior data sequence is encoded. Time dependence means that the occurrence of one competitive action may affect the selection and execution of the next competitive action. The purpose of the encoding process is to capture this time dependence and convert it into feature vectors. For example, the behavior data sequence can be encoded in a sequence. Considering the order and time interval of the actions, models such as recurrent neural networks or long short-term memory networks can be used to encode the sequence data to obtain the feature vectors of the behavior data sequence. These feature vectors indicate indicators such as the action type, action frequency, action coherence, and action success rate of each competitive action.

[0154] Then, based on the feature vector, evaluate the competitive style of the first virtual object. Before the evaluation, a set of competitive evaluation rules need to be predefined. For example, for the shooting percentage, it can be divided into three levels: low, medium, and high. Among them, below 35% is low, 36 - 55% is medium, and above 56% is high; for the number of steals, it can be divided into three levels: few, medium, and many. Among them, less than 1 steal in a single game is few, 2 - 3 steals is medium, and more than 4 steals is many. When evaluating the competitive style of the first virtual object based on the feature vector, the various indicators in the feature vector can be analyzed based on the competitive evaluation rules to understand the performance of the first virtual object in different competitive actions. For example, by comparing the data in the feature vector with the competitive evaluation rules, it can be judged whether the performance of the first virtual object in each aspect conforms to the characteristics of a certain competitive style; according to the application results of the rules, generate the evaluation results of the competitive style corresponding to the feature vector. This evaluation result may be a specific competitive style label or an evaluation report containing multiple dimensions. After obtaining the evaluation results of the competitive style, these evaluation results need to be converted into behavioral characteristics representing the competitive style of the first virtual object. This step involves matching the evaluation results with predefined standard features (i.e., the features defined in the competitive evaluation rules. For example, for the shooting percentage, it can be divided into three levels: low, medium, and high. Among them, below 35% is low, 36 - 55% is medium, and above 56% is high).

[0155] By matching the various indicators in the evaluation results with the standard features, find the standard features that best match the performance of the first virtual object, that is, use the matched standard features as the behavioral characteristics representing the competitive style of the first virtual object. For example, if the evaluation results show that the shooting percentage of the first virtual object is 60%, then "high shooting percentage" can be used as its behavioral characteristic. If the evaluation results show that the number of steals of the first virtual object in a single game is 6, then "many steals" can be used as its behavioral characteristic.

[0156] Through the above method, the behavioral characteristics of the first virtual object in multiple aspects can be obtained, and these behavioral characteristics are used for subsequent matching of training templates.

[0157] Step 203: Match the star template.

[0158] See Figure 6 , Figure 6 which is the schematic diagram of the star template provided by the embodiment of the present application. As Figure 6As shown, the star template library contains the key features of typical stars (i.e., the above-mentioned reference objects) and corresponding training strategies. Each typical star corresponds to multiple key features. For the convenience of subsequent matching descriptions, symbols can be used to replace the key features of typical stars. For example, the key features represented by the symbols include: A1 (a master at stealing the ball, with more than 4 steals in a single game), A2 (a master at shooting, with a high shooting percentage in a single game, exceeding 50% and more than 10 shots), A3 (the assist king, good at assisting, with more than 5 assists in a single game), A4 (a rebounding enthusiast, with more than 7 rebounds in a single game), A5 (a turnover master, with more than 5 turnovers in a single game), A6 (a three-point shooter, with a three-point shooting percentage exceeding 35% and making 3 three-pointers), A7 (a mid-range shooter, with a mid-range shooting percentage as high as 50% and making 4 mid-range shots), A8 (a dominant inside player, scoring 4 goals inside the paint in a single game), A9 (a marathon runner, with a running distance of more than 5 kilometers throughout the game), A10 (basketball IQ, running rationality, probability of open space appearance, and being able to create 3 open spaces in a single game). By way of example, the key features of star A are A1+A2+A3, that is, the key features of star A are a master at stealing the ball + a master at shooting + the assist king, and its corresponding training strategy (i.e., the above-mentioned training strategy) is B1; the key features of star B are A1+A3+A5, that is, the key features of star B are a master at stealing the ball + the assist king + a turnover master, and its corresponding training strategy is B2.

[0159] It can be understood that the above symbols and key features are only examples. In addition, there can be other types of key features and symbols. The embodiments of the present application do not limit the types of symbols and key features.

[0160] When matching the behavioral characteristics of the first virtual object with typical stars, assuming that the behavioral characteristics of the first virtual object are A1+A2+A3, the behavioral characteristics of the first virtual object can be matched with the key features of each typical star in the star template. The following formula can be used to calculate the probability of matching typical stars. The typical star with the highest probability is the star that matches the first virtual object, and the training strategy corresponding to the typical star with the highest probability is used as the training strategy of the first virtual object.

[0161] F(A) = A1*a1 + A2*a2 + A3*a3;

[0162] Among them, A represents the total sum of the characteristics of the first virtual object collected, which represents the total amount of all characteristics added together. A1, A2, and A3 represent the behavioral characteristics of the first virtual object. As shown above, A1 represents the characteristic of being a master at stealing the ball, A2 represents the characteristic of being a master at shooting, and A3 represents the characteristic of being good at assisting. a1, a2, and a3 represent the percentages of the corresponding key features of the stars in the star template library, and a1 + a2 + a3 = 100%.

[0163] See Figure 7 , Figure 7 Figure 7 is a matching schematic diagram of the star player template provided by the embodiments of the present application. Assume that the behavioral characteristics of the first virtual object are A1+A2+A3. There are 4 similar star players in the star player template library with these characteristics of A1+A2+A3: Star player A (A1+A2+A3), Star player E (A1+A2+A3+A7+A9), Star player F (A1+A2+A3+A4), Star player G (A1+A2+A3+A9); then, when matching the behavioral characteristics (A1+A2+A3) of the first virtual object with the star player template, in star player A, a1+a2+a3 = 100% (that is, the similarity between the first virtual object and star player A is 100%), and on average, a1 = a2 = a3 = 33.3%. In star player E, a1+a2+a3 = 60% because they account for 3 / 5 of the characteristics of star player E (that is, the similarity between the first virtual object and star player E is 60%), and on average, a1 = a2 = a3 = 20%. In star player F, a1+a2+a3 = 75% (that is, the similarity between the first virtual object and star player F is 75%), and on average, a1 = a2 = a3 = 25%. In star player G, a1+a2+a3 = 75% (that is, the similarity between the first virtual object and star player F is 75%), and on average, a1 = a2 = a3 = 25%. When finally 4 matching star player templates appear, the value of each star player is calculated through the formula F(A)=a1(A1+A2+A3). Among them, the larger the value of a1, the larger the whole value, and the closer the first virtual object is to this star player template. It can be seen that the first virtual object is most similar to star player A, so the training strategy B1 corresponding to star player A is used as the training strategy of the first virtual object.

[0164] Step 204: Behavioral pattern analysis.

[0165] See Figure 8 , Figure 8 Figure 8 is an analysis schematic diagram of the training strategy provided by the embodiments of the present application. Each typical star player (or star player model) in the star player template has corresponding key characteristics and training strategies, and each training strategy corresponds to corresponding advantage and disadvantage analyses. After determining the training strategy of the typical star player matched by the first virtual object, this training strategy can be output for the player to select.

[0166] In addition, time series analysis can also be used to detect the development trend of the player (i.e., the first virtual object) over time. The behavioral data of the player will be accumulated and compared. The more games the player plays, the more accurate the system's analysis of his behavior will be, so that the most suitable training strategy for the player can be found.

[0167] See Figure 9 , Figure 9It is a schematic diagram of the style analysis report provided by the embodiments of the present application. In actual applications, there are many types of star players. For example, in (a), if the typical star player matching the first virtual object is star player A, whose corresponding competitive style is a three-point finisher, the advantage is the ability to have stable scoring output; the weakness is that the footwork is relatively slow and one needs to have one's own rhythm to score relatively highly; the way to improve (enhance) is how to strengthen the offense. In (b), if the typical star player matching the first virtual object is star player B, whose corresponding competitive style is a team commander, the advantages are having a strong and tall body, being a great interior defender, having excellent rebounding and offensive capabilities, and also having good three-point and passing abilities; the weakness is that the footwork will be relatively slow and it is difficult to defend effectively; the way to improve (enhance) is to utilize one's own advantages in offense and let teammates make up for the deficiencies. In (c), if the typical star player matching the first virtual object is star player C, whose corresponding competitive style is a rim destroyer, the advantages are having powerful dunking and layup abilities, excellent physical explosiveness, outstanding offensive capabilities, and also having good shooting abilities; the weakness is that the defense is average and there may not be many offensive means; the way to improve (enhance) is to find the right opportunity to dunk and score in one step.

[0168] Step 205: Customize and generate a training plan.

[0169] Based on the above analysis results, design exclusive practice tasks for players. In this stage, the reinforcement learning strategy can be combined to dynamically adjust the task difficulty according to real-time feedback, making it both challenging and able to effectively help improve skills. The training tasks include the training tasks corresponding to the star player templates matched by the players and the training tasks designed specifically for the players.

[0170] Carry on Figure 9 the example in Figure 10 , Figure 10It is a schematic diagram of a training plan provided by an embodiment of the present application. In (a), if the typical star player matching the first virtual object is star player A and his corresponding competitive style is three-point finisher, then based on star player A, players can better adapt to the playing style of star player A and conduct specialized training. The training strategy of the three-point finisher mainly focuses on three-point offense and includes four aspects: training on shooting timing, training on getting rid of defenders, training on using pick-and-rolls, and exclusive training for star player A. Among them, the training on shooting timing mainly involves methods such as open-shot shooting and looking for open-shot opportunities. The training on getting rid of defenders is carried out using one's own offensive dribbling. The training on using pick-and-rolls is carried out using the running positions and cooperation of teammates. The training on the characteristic playing style of star player A includes the practice of star player A's dribbling + fakes + shooting. In (b), if the typical star player matching the first virtual object is star player B and his corresponding competitive style is team commander, the training strategy of the team commander mainly focuses on the post play of the big center, mainly on offense and linking up the whole team. His training includes training on passing timing, training on inside offensive footwork, training on running positions after pick-and-rolls, and exclusive training for star player B. In (c), if the typical star player matching the first virtual object is star player C and his corresponding competitive style is rim protector, the training strategy of the rim protector mainly focuses on breakthrough offense and includes training on dunk timing, training on looking for dunk opportunities, training on using pick-and-rolls, and exclusive training for star player C.

[0171] When players obtain targeted training strategies, they can start targeted training, which is convenient for players to better familiarize themselves with the game, the star players, and the playing styles, and can understand their own styles, so as to better play to their own and the star players' advantages to win for the team.

[0172] Step 206: Adjust the training plan.

[0173] In practical applications, a feedback channel can be provided to allow users to feedback their subjective evaluations of the effects of training strategies. Such opinions will also be incorporated into the template update process to continuously iterate and update the star player data and star player characteristics in the star player template library to make them more in line with individual needs.

[0174] Through the above method, when players play with their favorite star players or star players with similar styles and don't know how to bring out their best abilities, a highly customized learning method and content are given to players. Such a customized learning assistant can help players learn, find a style route suitable for themselves, and thus find out what kind of star player playing style suits them better to understand the game and have fun playing the game.

[0175] So far, the processing method of the training strategy provided in the embodiments of the present application has been described in combination with the exemplary applications and implementations of the electronic device provided in the embodiments of the present application. Next, the cooperation of each module in the processing device 555 of the training strategy provided in the embodiments of the present application to implement the processing solution of the training strategy will be described.

[0176] The reference determination module 5551 is configured to, in response to a strategy customization operation for a first virtual object in a virtual scene corresponding to competitive sports, determine a first reference object that matches the first virtual object based on the behavior data of the first virtual object in the competitive interaction in the virtual scene; the strategy processing module 5552 is configured to generate a first training strategy adapted to the first virtual object based on the first reference object and output the first training strategy.

[0177] In some embodiments, the reference determination module is further configured to extract features from the behavior data of the first virtual object in the competitive interaction in the virtual scene to obtain behavior features characterizing the competitive style of the first virtual object; match the behavior features with the key features of multiple reference objects, and determine a first reference object that matches the first virtual object from the multiple reference objects based on the matching result.

[0178] In some embodiments, when the behavior data of the first virtual object in the competitive interaction in the virtual scene includes behavior data of multiple competitive actions, the reference determination module is further configured to perform normalization processing on the behavior data to obtain a behavior data sequence; perform encoding processing on the behavior data sequence based on the time dependence of each competitive action in the behavior data sequence to obtain a feature vector of the behavior data sequence, where the feature vector indicates at least one of the following action indicators of each competitive action: action type, action frequency, action coherence, action success rate; evaluate the competitive style of the first virtual object based on the feature vector, and determine the behavior features characterizing the competitive style of the first virtual object based on the evaluation result.

[0179] In some embodiments, the reference determination module is further configured to evaluate the competitive style of the first virtual object based on the competitive evaluation rules corresponding to the virtual scene and the feature vector to obtain an evaluation result of the competitive style corresponding to the feature vector; determine a standard feature corresponding to the competitive style of the feature vector, and use the standard feature as the behavior feature characterizing the competitive style of the first virtual object.

[0180] In some embodiments, the reference determination module is further configured to calculate the similarity between the behavior characteristics and the key characteristics of each reference object, and use the similarity between the behavior characteristics and the key characteristics of each reference object as the matching result; screen out the reference object corresponding to the maximum similarity among the multiple reference objects, and use the screened reference object as the first reference object that matches the first virtual object.

[0181] In some embodiments, the policy processing module is further configured to generate a reference training policy for the first reference object based on the competitive style of the first reference object, and use the reference training policy as the first training policy adapted to the first virtual object.

[0182] In some embodiments, the policy processing module is further configured to determine the competitive disadvantage of the first virtual object based on the behavior characteristics of the first virtual object, and generate a strengthening training policy for overcoming the competitive disadvantage based on the competitive disadvantage; determine the reference training policy of the first reference object, and fuse the reference training policy and the strengthening training policy to obtain the first training policy adapted to the first virtual object.

[0183] In some embodiments, the training policy customization operation is triggered by triggering a policy customization entry, and the device further includes: an entry display module configured to perform at least one of the following: display the policy customization entry when the competitive interaction of the first virtual object ends; display the policy customization entry in the cultivation training interface of the first virtual object; display the policy customization entry during the competitive interaction of the first virtual object.

[0184] In some embodiments, after determining the first reference object that matches the first virtual object, the device further includes: a competitive style display module configured to display an analysis interface for the competitive style of the first virtual object based on the first reference object; display the competitive style of the first virtual object, the first reference object that matches the first virtual object, the competitive advantages and disadvantages of the first reference object, and training tips for overcoming the competitive disadvantages in the competitive style analysis interface.

[0185] In some embodiments, the policy processing module is further configured to display a training tip interface for the first virtual object; in response to the first training policy including sub-training policies for multiple training items, display training entries for each training item in the training tip interface; and output the sub-training policy of the training item corresponding to the selected training entry in response to a selection operation.

[0186] In some embodiments, the policy processing module is further configured to determine impact factors for affecting the training priorities of the training items, where the impact factors include at least one of the following: required duration, difficulty, type, and synergistic effect on other training items; predict the training priorities of the training items based on the impact factors, and display the training entrances of the training items in the training prompt interface in a manner corresponding to the training priorities.

[0187] In some embodiments, the policy processing module is further configured to output the first training policy in the form of policy prompt information in response to meeting the policy prompt condition; where the policy prompt information includes the target competitive action recommended to be executed based on the first training policy and the timing of executing the target competitive action.

[0188] In some embodiments, the device further includes: a condition determination module, configured to determine that the policy prompt condition is met in response to a trigger operation on the policy prompt entrance when the policy prompt entrance is displayed in the interface of the virtual scene; or determine that the policy prompt condition is met in response to the interaction progress of the first virtual object in the competitive interaction reaching a target progress, where the target progress includes at least one of the following: there is an association between the competitive action currently executed by the first virtual object and the target competitive action recommended to be executed by the first training policy, and the competitive disadvantage targeted by the first training policy appears on the first virtual object.

[0189] In some embodiments, after outputting the first training policy, the device further includes: a policy update module, configured to determine the improvement degree for each execution of the first training policy against the competitive disadvantage when the first virtual object executes the first training policy multiple times to overcome the competitive disadvantage; fuse the improvement degrees for each execution of the first training policy against the competitive disadvantage to obtain a comprehensive improvement degree; and in response to the comprehensive improvement degree being lower than the degree threshold, replace the first training policy with a second training policy, where the second training policy is generated based on the skill style of a second reference object, and the similarity between the behavioral characteristics of the first virtual object and the key characteristics of the second reference object is the same as the similarity between the behavioral characteristics of the first virtual object and the key characteristics of the first reference object.

[0190] In some embodiments, after outputting the first training strategy, the policy update module is further configured to, in response to an operation of using the first training strategy, control the first virtual object to execute the first training strategy in the competitive interaction; during the process of the first virtual object executing the first training strategy, detect the competitive style of the first virtual object based on new behavior data; and in the case where there is a difference between the current competitive style and the original competitive style, adjust the first training strategy so that the adjusted first training strategy is adapted to the current competitive style.

[0191] In some embodiments, after outputting the first training strategy, the policy update module is further configured to, in response to a feedback operation, obtain feedback information regarding the first training strategy; and update and adjust the first training strategy based on the feedback information.

[0192] In some embodiments, after outputting the first training strategy, the apparatus further includes: a policy comparison module, configured to create a training challenge task for the first training strategy, and send an invitation request for the training challenge task to the terminal of a second virtual object, where the second virtual object is matched with the first target object; and in response to the second virtual object agreeing to the invitation request and joining the training challenge task, compare the training result of the first virtual object for the first training strategy with the training result of the second virtual object for the first training strategy.

[0193] 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 processing method of the training strategy in the embodiments of the present application as described above.

[0194] An embodiment of the present application provides a computer-readable storage medium storing computer executable instructions, where computer executable instructions or a computer program are stored therein, and when the computer executable instructions or the computer program are executed by a processor, the processor will be caused to execute the processing method of the training strategy provided by the embodiment of the present application. For example, as Figure 3 shown in the processing method of the training strategy.

[0195] 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; or may be various devices including one or any combination of the above memories.

[0196] In some embodiments, the computer-executable instructions may be in the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including being deployed as a stand-alone program or being deployed as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0197] As an example, the computer-executable instructions may or may not correspond to a file in a file system, may be stored as part of a file that holds other programs or data, for example, 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 (e.g., files that store one or more modules, subroutines, or portions of code).

[0198] As an example, the computer-executable instructions may 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 by a communication network.

[0199] As described above, the above are only embodiments of the present application and are not intended to limit the protection scope of the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and scope of the present application are included within the protection scope of the present application.

Claims

1. A method for processing a training strategy, characterized in that: The method comprises: In response to a training strategy customization operation for a first virtual object in a virtual scene corresponding to competitive sports, determining a first reference object that matches the first virtual object based on the behavioral data of the first virtual object's competitive interaction in the virtual scene; A first training strategy adapted to the first virtual object is obtained according to the first reference object, and the first training strategy is output.

2. The method according to claim 1, characterized in that The determining a first reference object matching the first virtual object based on the behavior data of the first virtual object's competitive interaction in the virtual scene includes: Extracting features from the behavioral data of the first virtual object's competitive interaction in the virtual scene to obtain behavioral features that characterize the competitive style of the first virtual object; The behavior feature is matched with key features of a plurality of reference objects, and a first reference object matching the first virtual object is determined from among the plurality of reference objects based on the matching result.

3. The method according to claim 2, characterized in that The extracting features of the behavior data of the first virtual object's competitive interaction in the virtual scene to obtain the behavior features representing the competitive style of the first virtual object includes: When the behavior data of the competitive interaction of the first virtual object in the virtual scene includes behavior data of a plurality of competitive actions, performing standardization processing on the behavior data to obtain a behavior data sequence; Based on the time dependency of each of the competitive actions in the behavior data sequence, the behavior data sequence is encoded to obtain a feature vector of the behavior data sequence, wherein the feature vector indicates at least one of the following action indicators of each of the competitive actions: action type, action frequency, action continuity, and action success rate; The competitive style of the first virtual object is evaluated based on the feature vector, and a behavior feature characterizing the competitive style of the first virtual object is determined based on the evaluation result.

4. The method according to claim 3, characterized in that The evaluating the competitive style of the first virtual object based on the feature vector includes: Based on the competition evaluation rule corresponding to the virtual scene and the feature vector, the competition style of the first virtual object is evaluated to obtain an evaluation result of the competition style corresponding to the feature vector; The determining, based on the evaluation result, a behavior feature characterizing the competitive style of the first virtual object comprises: A standard feature corresponding to the competitive style of the feature vector is determined, and the standard feature is used as a behavior feature characterizing the competitive style of the first virtual object.

5. The method according to claim 2, characterized in that: The matching of the behavior feature with key features of multiple reference objects includes: Calculating similarity between the behavior feature and the key feature of each reference object, and taking the similarity between the behavior feature and the key feature of each reference object as the matching result; The determining, based on the matching result, a first reference object matching the first virtual object from among the plurality of reference objects comprises: A reference object corresponding to the greatest similarity is screened out from the multiple reference objects, and the screened out reference object is used as a first reference object that matches the first virtual object.

6. The method according to claim 1, characterized in that The step of obtaining a first training strategy adapted to the first virtual object according to the first reference object comprises: A reference training strategy of the first reference object is acquired, and the reference training strategy is used as a first training strategy adapted to the first virtual object.

7. The method according to claim 1, characterized in that The step of obtaining a first training strategy adapted to the first virtual object according to the first reference object comprises: Determining a competitive disadvantage of the first virtual object based on the behavioral characteristics of the first virtual object, and generating an enhanced training strategy for overcoming the competitive disadvantage based on the competitive disadvantage; A reference training strategy for the first reference object is determined, and the reference training strategy and the enhanced training strategy are merged to obtain the first training strategy adapted to the first virtual object.

8. The method according to claim 1, characterized in that The training strategy customization operation is triggered by triggering a strategy customization portal, and the display mode of the strategy customization portal includes at least one of: When the competitive interaction of the first virtual object ends, displaying a strategy customization entrance; In the training interface for developing the first virtual object, a strategy customization entrance is displayed; During the process of the first virtual object performing competitive interaction, a strategy customization entrance is displayed.

9. The method according to claim 1, characterized in that: After determining the first reference object matching the first virtual object, the method further includes: displaying a competition style analysis interface of the first virtual object based on the first reference object; The competitive style of the first virtual object, a first reference object matching the first virtual object, competitive advantages and disadvantages of the first reference object, and training tips for overcoming the competitive disadvantages are displayed in the competitive style analysis interface.

10. The method according to claim 1, characterized in that The outputting the first training strategy includes: Displaying a training prompt interface for the first virtual object; In response to the first training strategy including sub-training strategies for a plurality of training items, displaying a training entry for each of the training items in the training prompt interface; In response to the selection operation, the sub-training strategy of the training item corresponding to the selected training entry is output.

11. The method according to claim 10, characterized in that The training entry for each training item is displayed in the training prompt interface, including: Determine an influencing factor for influencing the training priority of the training project, the influencing factor comprising at least one of the following: required duration, difficulty, type, and synergistic effect on other training projects; The training priority of each of the training items is predicted based on the influencing factors, and the training entry of each of the training items is displayed in the training prompt interface in a manner corresponding to the training priority.

12. The method according to claim 1, characterized in that The first training strategy for outputting the first virtual object includes: In response to satisfying the strategy prompt condition, outputting the first training strategy in the form of strategy prompt information; The strategy prompt information includes a target competitive action recommended for execution based on the first training strategy and a timing for executing the target competitive action.

13. The method according to claim 12, characterized in that The method further comprises: In the case where a strategy prompt entry is displayed in the interface of the virtual scene, in response to a trigger operation on the strategy prompt entry, determining that a strategy prompt condition is satisfied; or, In response to the interaction progress of the first virtual object in competitive interaction reaching the target progress, it is determined that the strategy prompt condition is met, and the target progress includes at least one of the following: the competitive action currently performed by the first virtual object is associated with the target competitive action recommended to be performed by the first training strategy, and the competitive disadvantage targeted by the first training strategy appears on the first virtual object.

14. The method according to claim 1, characterized in that After outputting the first training strategy, the method further includes: When the first virtual object executes the first training strategy multiple times to overcome a competitive disadvantage, determining a degree of improvement of the competitive disadvantage each time the first training strategy is executed; The improvement degree of the competitive disadvantage obtained by executing the first training strategy each time is integrated to obtain a comprehensive improvement degree; In response to the comprehensive improvement degree being lower than a degree threshold, the first training strategy is replaced with a second training strategy, wherein the second training strategy is generated based on the skill style of a second reference object, and the similarity between the behavioral characteristics of the first virtual object and the key characteristics of the second reference object is the same as the similarity between the behavioral characteristics of the first virtual object and the key characteristics of the first reference object.

15. The method according to claim 1, characterized in that After outputting the first training strategy, the method further includes: In response to a use operation on the first training strategy, controlling the first virtual object to execute the first training strategy in a competitive interaction; During the process in which the first virtual object executes the first training strategy, detecting the competitive style of the first virtual object based on the new behavior data; When there is a difference between the current competitive style and the original competitive style, the first training strategy is adjusted so that the adjusted first training strategy is adapted to the current competitive style.

16. The method according to claim 1, characterized in that After outputting the first training strategy, the method further includes: In response to a feedback operation, obtaining feedback information for the first training strategy; The first training strategy is updated and adjusted based on the feedback information.

17. The method according to claim 1, after outputting the first training strategy, the method further comprises: Creating a training challenge task for the first training strategy, and publishing the training challenge task to a terminal of a second virtual object, the second virtual object matching the first reference object; In response to the second virtual object accepting the training challenge task, a training result of the first virtual object for the first training strategy is compared with a training result of the second virtual object for the first training strategy.

18. A training strategy processing device, characterized in that: The device comprises: A reference determination module, configured to determine, in response to a strategy customization operation for a first virtual object in a virtual scene corresponding to competitive sports, a first reference object matching the first virtual object based on the behavior data of the first virtual object's competitive interaction in the virtual scene; A strategy processing module is used to generate a first training strategy adapted to the first virtual object based on the first reference object, and output the first training strategy.

19. An electronic device, characterized in that: include: A memory for storing computer executable instructions or computer programs; A processor, used to implement the processing method of the training strategy described in any one of claims 1 to 17 when executing the computer executable instructions or computer program stored in the memory.

20. A computer-readable storage medium, characterized in that: Computer executable instructions or computer programs are stored, and when the computer executable instructions or computer programs are executed by a processor, the processing method of the training strategy described in any one of claims 1 to 17 is implemented.

21. A computer program product comprising a computer program or computer executable instructions, characterized in that: When the computer program or computer executable instructions are executed by a processor, the processing method of the training strategy described in any one of claims 1 to 17 is implemented.

Images