Service attribute updating method and device, equipment and storage medium

Abstract

The invention relates to a service attribute updating method and device, equipment and a storage medium. The method comprises the steps of obtaining a lineup attribute information set corresponding to each seed lineup in a seed lineup pool of a preset service; for the array capacity attribute information set corresponding to each seed array capacity, dividing the array capacity attribute information set into front-row array capacity attribute information and rear-row array capacity attribute information according to an attribute type corresponding to each piece of array capacity attribute information; sequentially taking the front-row array capacity attribute information or the rear-row array capacity attribute information corresponding to each seed array capacity as fixed attribute information of each seed array capacity; fixing the fixed attribute information of each seed lineup in the seed lineup pool, and iteratively updating the non-fixed attribute information of each seed lineup to obtain an updated lineup pool; and determining a target lineup based on a battle result between lineups in the updated lineup pool. The probability that the screened target lineup is the high-quality lineup is improved.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of computer, and particularly relates to a service attribute updating method and device, equipment and a storage medium. BACKGROUND

[0002] In the related art, in the field of electronic game development, game balance testing is a key link to ensure game playability and long-term operation success. With the diversification of game types and the complication of game mechanisms, traditional balance testing methods face many challenges. The current mainstream testing methods mainly include three categories: manual testing, A / B testing and automated simulation testing.

[0003] The manual testing method mainly relies on the experience judgment of game planners, and adjusts parameters and simulates battles to evaluate balance. Although this method is intuitive, it is inefficient and difficult to cover all possible strategy combinations in the game. A / B testing adjusts game balance by collecting real player behavior data, although the results are reliable, the testing period is long and may affect the player experience. Automated simulation testing uses statistical simulation methods to improve testing efficiency, but the current AI intelligence level and behavior patterns are still significantly different from real players, which may lead to testing results deviating from real battle situations, making it difficult to accurately recommend high-quality lineups to users. SUMMARY

[0004] The present disclosure provides a service attribute updating method, device, equipment and storage medium to at least solve the problem of high difficulty in screening high-quality lineups in the related art. The technical solutions of the present disclosure are as follows: According to a first aspect of the embodiments of the present disclosure, a service attribute updating method is provided, comprising: obtaining a set of lineup attribute information corresponding to each seed lineup in a seed lineup pool of a preset service; the set of lineup attribute information includes at least two sets of lineup attribute information; For each set of lineup attribute information corresponding to each seed lineup, the set of lineup attribute information is divided into front-row lineup attribute information and back-row lineup attribute information according to the attribute type corresponding to each set of lineup attribute information, to obtain the front-row lineup attribute information and the back-row lineup attribute information corresponding to each seed lineup; sequentially taking the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup as the fixed attribute information of each seed lineup; fixing the fixed attribute information of each seed lineup in the seed lineup pool, and iteratively updating the non-fixed attribute information of each seed lineup to obtain an updated lineup pool; the non-fixed attribute information of each seed lineup is the attribute information other than the fixed attribute information in the front-row lineup attribute information and the back-row lineup attribute information corresponding to each seed lineup; Determine a target lineup based on the results of battles between lineups in the updated lineup pool.

[0005] In an exemplary embodiment, the front-line lineup attribute information or the back-line lineup attribute information corresponding to each seed lineup is taken as fixed attribute information of each seed lineup; the fixed attribute information of each seed lineup in the seed lineup pool is fixed, and the non-fixed attribute information of each seed lineup is iteratively updated to obtain an updated lineup pool, including: The front-line lineup attribute information or the back-line lineup attribute information corresponding to each seed lineup is taken as first lineup attribute information of each seed lineup; and attribute information other than the first lineup attribute information in the front-line lineup attribute information or the back-line lineup attribute information corresponding to each seed lineup is determined as second lineup attribute information of each seed lineup; The second lineup attribute information of each seed lineup in the seed lineup pool is fixed, and the first lineup attribute information of each seed lineup is iteratively updated to obtain an initial updated lineup pool; the initial updated lineup pool includes the second lineup attribute information and first updated lineup attribute information corresponding to each initial updated lineup; the first updated lineup attribute information is attribute information updated from the first lineup attribute information; The updated lineup pool is constructed based on the results of battles between initial updated lineups in the initial updated lineup pool and seed lineups in the seed lineup pool.

[0006] In an exemplary embodiment, the updated lineup pool is constructed based on the results of battles between initial updated lineups in the initial updated lineup pool and seed lineups in the seed lineup pool, including: Obtain parent-child lineup battle results between each initial updated lineup in the initial updated lineup pool and each seed lineup in the seed lineup pool; Determine a screening lineup pool from the initial updated lineup pool and the seed lineup pool according to the battle results; The first lineup attribute information of each screening lineup in the screening lineup pool is fixed, and the second lineup attribute information of each screening lineup is iteratively updated to obtain second updated lineup attribute information; An updated lineup is constructed based on the first lineup attribute information and the second updated lineup attribute information corresponding to each screening lineup, and the updated lineup pool is constructed based on the updated lineup.

[0007] In an exemplary embodiment, the second lineup attribute information of each seed lineup in the seed lineup pool is fixed, and the first lineup attribute information of each seed lineup is iteratively updated to obtain an initial updated lineup pool, including: The second lineup attribute information of each seed lineup in the seed lineup pool is fixed, and a lineup crossover operation is performed based on the first lineup attribute information of each seed lineup to obtain a first updated lineup; A lineup variation operation is performed based on the first lineup attribute information of each seed lineup to obtain a second updated lineup; Based on the first updated lineup and the second updated lineup, the initial updated lineup pool is constructed.

[0008] In an exemplary embodiment, the acquisition of the parent-child lineup battle result between each initial updated lineup in the initial updated lineup pool and each seed lineup in the seed lineup pool includes: Each initial updated lineup in the initial updated lineup pool is acquired as a parent lineup, and a seed lineup in the seed lineup pool is acquired as a child lineup; The parent-child lineup battle result is determined according to the win rate, survival rate, and battle loss ratio of the parent lineup and the child lineup in a preset business scenario.

[0009] In an exemplary embodiment, the determination of the parent-child lineup battle result according to the win rate, survival rate, and battle loss ratio of the parent lineup and the child lineup in a preset business scenario includes: The win rate, survival rate, and battle loss ratio of the parent lineup and the child lineup in a round-robin scenario are acquired to obtain a first battle result; The win rate, survival rate, and battle loss ratio of the parent lineup and the child lineup in a tournament scenario are acquired to obtain a second battle result; The parent-child lineup battle result is determined according to the first battle result and the second battle result.

[0010] In an exemplary embodiment, the determination of the screening lineup from the initial updated lineup pool and the seed lineup pool according to the battle result includes: The role similarity, skill similarity, and formation similarity between the parent lineups in the seed lineup pool are determined according to the lineup attribute information corresponding to the parent lineups in the seed lineup pool; The lineup difference result of the seed lineup pool is determined according to the role similarity, skill similarity, and formation similarity between the parent lineups; The role similarity, skill similarity, and formation similarity between the child lineups in the initial updated lineup pool are determined according to the lineup attribute information corresponding to the child lineups in the initial updated lineup pool; The lineup difference result of the initial updated lineup pool is determined according to the role similarity, skill similarity, and formation similarity between the child lineups; determining a screening lineup from the initial updated lineup pool and the seed lineup pool based on the competition result, the lineup difference result of the seed lineup pool and the lineup difference result of the initial updated lineup pool.

[0011] In an exemplary embodiment, the determining a screening lineup from the initial updated lineup pool and the seed lineup pool based on the competition result, the lineup difference result of the seed lineup pool and the lineup difference result of the initial updated lineup pool comprises: determining a first competition score of the seed lineup pool and a second competition score of the initial updated lineup pool based on the competition result; determining a comprehensive evaluation score of the seed lineup pool based on the first competition score and the lineup difference result of the seed lineup pool; determining a comprehensive evaluation score of the initial updated lineup pool based on the second competition score and the lineup difference result of the initial updated lineup pool; determining a screening lineup from the initial updated lineup pool and the seed lineup pool according to the comprehensive evaluation score of the seed lineup pool and the comprehensive evaluation score of the initial updated lineup pool.

[0012] In an exemplary embodiment, the determining the lineup difference result of the seed lineup pool according to the role similarity, the skill similarity and the formation similarity between the parent lineups comprises: determining a first difference result according to the role similarity between the parent lineups; determining a second difference result according to the skill similarity between the parent lineups; determining a third difference result according to the formation similarity between the parent lineups; obtaining a first weight corresponding to the role similarity, a second weight corresponding to the skill similarity and a third weight corresponding to the formation similarity; determining the lineup difference result of the seed lineup pool according to the product of the first difference result and the first weight, the product of the second difference result and the second weight and the product of the third difference result and the third weight.

[0013] In an exemplary embodiment, the constructing the updated lineup pool based on the competition result between the initial updated lineup in the initial updated lineup pool and the seed lineup in the seed lineup pool comprises: obtaining the competition result between each initial updated lineup in the initial updated lineup pool; screening out candidate lineups from the initial updated lineup pool according to the competition result between each initial updated lineup to construct a candidate lineup pool; According to a battle result between the candidate lineup pool and the seed lineup in the seed lineup pool, an updated lineup pool is constructed by using a restricted tournament replacement strategy.

[0014] In an exemplary embodiment, the determining the target lineup based on the battle results between the lineups in the updated lineup pool comprises: According to the updated lineup pool, at least two lineup combinations are constructed; A respective battle result of each lineup combination in at least two preset business scenarios is obtained; According to the respective battle results of each lineup combination in the at least two preset business scenarios, the target lineup is filtered out from the updated lineup pool; The method further comprises: The target lineup attribute information is sent to a terminal, so that the terminal displays the target lineup attribute information.

[0015] According to a second aspect of the embodiments of the present disclosure, a business attribute updating apparatus is provided, comprising: A lineup attribute acquisition module configured to perform acquisition of a set of lineup attribute information corresponding to each seed lineup in a seed lineup pool of a preset business; the set of lineup attribute information comprises at least two pieces of lineup attribute information; A lineup attribute classification module configured to perform, for the set of lineup attribute information corresponding to each seed lineup, division of the set of lineup attribute information into front-row lineup attribute information and back-row lineup attribute information according to an attribute type corresponding to each piece of lineup attribute information, to obtain the front-row lineup attribute information and the back-row lineup attribute information corresponding to each seed lineup; A lineup pool updating module configured to perform, in sequence, the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup as fixed attribute information of each seed lineup; fixing the fixed attribute information of each seed lineup in the seed lineup pool, and iteratively updating non-fixed attribute information of each seed lineup to obtain an updated lineup pool; the non-fixed attribute information of each seed lineup is attribute information other than the fixed attribute information in the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup; A target lineup determination module configured to perform determination of a target lineup based on a battle result between lineups in the updated lineup pool.

[0016] In an exemplary embodiment, the lineup pool updating module comprises: The attribute classification unit is configured to perform the following: taking the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup as first lineup attribute information of each seed lineup; and determining attribute information other than the first lineup attribute information from the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup as second lineup attribute information of each seed lineup. The initial updating unit is configured to perform the following: fixing the second lineup attribute information of each seed lineup in the seed lineup pool, and iteratively updating the first lineup attribute information of each seed lineup to obtain an initial updated lineup pool; the initial updated lineup pool includes the second lineup attribute information and first updated lineup attribute information corresponding to each initial updated lineup; the first updated lineup attribute information is attribute information updated from the first lineup attribute information. The lineup pool construction unit is configured to perform the following: constructing the updated lineup pool based on a battle result between an initial updated lineup in the initial updated lineup pool and a seed lineup in the seed lineup pool.

[0017] In an exemplary embodiment, the lineup pool construction unit includes: The battle result acquisition subunit is configured to perform the following: acquiring a parent-child lineup battle result between each initial updated lineup in the initial updated lineup pool and each seed lineup in the seed lineup pool. The lineup screening subunit is configured to perform the following: determining a screening lineup pool from the initial updated lineup pool and the seed lineup pool according to the battle result. The iteration subunit is configured to perform the following: fixing the first lineup attribute information of each screening lineup in the screening lineup pool, and iteratively updating the second lineup attribute information of each screening lineup to obtain second updated lineup attribute information. The lineup construction subunit is configured to perform the following: constructing an updated lineup based on the first lineup attribute information and the second updated lineup attribute information corresponding to each screening lineup, and constructing the updated lineup pool based on the updated lineup; if the screening lineup is the initial updated lineup, the first lineup attribute information of the screening lineup is the first updated lineup attribute information; if the screening lineup is the seed lineup, the first lineup attribute information of the screening lineup is the first lineup attribute information of the seed lineup.

[0018] In an exemplary embodiment, the iteration subunit comprises, and is further configured to perform: fixing the second lineup attribute information of each seed lineup in the seed lineup pool, performing a lineup crossover operation based on the first lineup attribute information of each seed lineup to obtain a first updated lineup, and performing a lineup mutation operation based on the first lineup attribute information of each seed lineup to obtain a second updated lineup, and constructing the initial updated lineup pool based on the first updated lineup and the second updated lineup.

[0019] In an exemplary embodiment, the battle result obtaining subunit comprises, and is further configured to perform: obtaining each initial updated lineup in the initial updated lineup pool as a parent lineup, and obtaining a seed lineup in the seed lineup pool as a child lineup, and determining the battle result of the parent-child lineups according to the win rate, survival rate and battle damage ratio of the parent-child lineups in a preset business scenario.

[0020] In an exemplary embodiment, the battle result obtaining subunit comprises, and is further configured to perform: obtaining the win rate, survival rate and battle damage ratio of the parent-child lineups in a round-robin scenario to obtain a first battle result, and obtaining the win rate, survival rate and battle damage ratio of the parent-child lineups in a tournament scenario to obtain a second battle result, and determining the battle result of the parent-child lineups according to the first battle result and the second battle result.

[0021] In an exemplary embodiment, the lineup screening subunit comprises: A first similarity determining subunit configured to determine the role similarity, skill similarity and formation similarity between the parent lineups in the seed lineup pool according to the lineup attribute information corresponding to the parent lineups; A difference result determining subunit configured to determine the lineup difference result of the seed lineup pool according to the role similarity, skill similarity and formation similarity between the parent lineups; A second similarity determining subunit configured to determine the role similarity, skill similarity and formation similarity between the child lineups in the initial updated lineup pool according to the lineup attribute information corresponding to the child lineups; A lineup difference determining subunit configured to determine the lineup difference result of the initial updated lineup pool according to the role similarity, skill similarity and formation similarity between the child lineups; A screening subunit configured to determine the screening lineup from the initial updated lineup pool and the seed lineup pool based on the battle result, the lineup difference result of the seed lineup pool and the lineup difference result of the initial updated lineup pool.

[0022] In one exemplary implementation, the filtering subunit is configured to perform the following actions: determining a first battle score for the seed lineup pool and a second battle score for the initial updated lineup pool based on the battle results; determining a comprehensive evaluation score for the seed lineup pool based on the first battle score and the lineup difference results of the seed lineup pool; determining a comprehensive evaluation score for the initial updated lineup pool based on the second battle score and the lineup difference results of the initial updated lineup pool; and determining a selection lineup from the initial updated lineup pool and the seed lineup pool based on the comprehensive evaluation scores of the seed lineup pool and the initial updated lineup pool.

[0023] In one exemplary implementation, the lineup difference determination subunit is configured to perform the following operations: determine a first difference result based on the character similarity between the parent lineups; determine a second difference result based on the skill similarity between the parent lineups; determine a third difference result based on the formation similarity between the parent lineups; obtain a first weight corresponding to the character similarity, a second weight corresponding to the skill similarity, and a third weight corresponding to the formation similarity; and determine the lineup difference result of the seed lineup pool based on the product of the first difference result and the first weight, the product of the second difference result and the second weight, and the product of the third difference result and the third weight.

[0024] In one exemplary implementation, the lineup pool construction unit includes: The battle result acquisition subunit is configured to acquire the battle results between each initial updated lineup in the initial updated lineup pool. The candidate lineup construction subunit is configured to perform the task of selecting candidate lineups from the initial updated lineup pool based on the battle results between each initial updated lineup, and constructing a candidate lineup pool. The lineup pool construction subunit is configured to construct the updated lineup pool based on the battle results between the candidate lineup pool and the seed lineups in the seed lineup pool, using a restricted tournament replacement strategy.

[0025] In one exemplary implementation, the target lineup determination module includes: The lineup combination building unit is configured to build at least two lineup combinations based on the updated lineup pool; The battle result acquisition unit is configured to acquire the battle results corresponding to each lineup combination in at least two preset business scenarios. The lineup selection unit is configured to select the target lineup from the updated lineup pool based on the battle results of each lineup combination in at least two preset business scenarios. The device further includes: The lineup attribute sending module is configured to send the lineup attribute information of the target lineup to the terminal so that the terminal can display the lineup attribute information of the target lineup.

[0026] According to a third aspect of the present disclosure, an electronic device is provided, comprising: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the business attribute update method described above.

[0027] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, which, when the instructions in the computer-readable storage medium are executed by an electronic device processor, enables the electronic device to perform the business attribute update method as described above.

[0028] According to a fifth aspect of the present disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the business attribute update method as described above.

[0029] The technical solutions provided by the embodiments of this disclosure have at least the following beneficial effects: This disclosure obtains a set of lineup attribute information corresponding to each seed lineup in a seed lineup pool for a preset service; the set of lineup attribute information includes at least two sets of lineup attribute information; for each set of lineup attribute information corresponding to a seed lineup, the set of lineup attribute information is divided into front-row lineup attribute information and back-row lineup attribute information according to the attribute type corresponding to each set of lineup attribute information, to obtain the front-row lineup attribute information and back-row lineup attribute information corresponding to each seed lineup; then, the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup is sequentially used as the fixed attribute information of each seed lineup; the fixed attribute information of each seed lineup in the seed lineup pool is fixed. The system collects and iteratively updates the non-fixed attribute information of each seed lineup to obtain an updated lineup pool. The non-fixed attribute information of each seed lineup includes the attribute information of the front-row lineup and the attribute information of the back-row lineup, excluding the fixed attribute information. By fixing the front-row or back-row attributes in the lineup, the search space is effectively limited to a few positions, significantly reducing the difficulty of searching for high-quality lineups and promoting the generation of diverse solutions within a single row domain. On the one hand, it significantly reduces the search space; on the other hand, searching within a single row domain (front-row lineup attribute information or back-row lineup attribute information) is more likely to produce diverse suboptimal results, ensuring lineup diversity and reducing the risk of premature convergence. Finally, based on the battle results between lineups in the updated lineup pool, the target lineup is determined. This can increase the probability that the selected target lineup is a high-quality lineup and increase the diversity of the selected target lineups.

[0030] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0031] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure, and are not intended to unduly limit this disclosure.

[0032] Figure 1 This is an application environment diagram illustrating a business attribute update method according to an exemplary embodiment.

[0033] Figure 2 This is a flowchart illustrating a business attribute update method according to an exemplary embodiment.

[0034] Figure 3 This is a flowchart illustrating a method for iteratively updating non-fixed attribute information of each seed lineup to obtain an updated lineup pool, according to an exemplary embodiment.

[0035] Figure 4This is a flowchart illustrating a method for determining a selected lineup from the initial updated lineup pool and the seed lineup pool based on the battle results, according to an exemplary embodiment.

[0036] Figure 5 This is a schematic diagram of a battle lineup in a full round-robin tournament, a non-full round-robin tournament, and a tournament scenario, according to an exemplary embodiment.

[0037] Figure 6 This is a flowchart illustrating a lineup update method in multiple scenarios according to an exemplary embodiment.

[0038] Figure 7 This is a schematic diagram illustrating a visualization of the Top roster in a round-robin tournament, according to an exemplary embodiment.

[0039] Figure 8 This is a schematic diagram illustrating the results of merging and observing search results from multiple main loops (T+x cycles) according to an exemplary embodiment.

[0040] Figure 9 This is a schematic diagram illustrating a seed pool update method according to an exemplary embodiment.

[0041] Figure 10 This is a block diagram illustrating a business attribute update apparatus according to an exemplary embodiment.

[0042] Figure 11 This is a block diagram illustrating a server according to an exemplary embodiment.

[0043] Figure 12 This is a block diagram illustrating an electronic device for updating business attributes according to an exemplary embodiment. Detailed Implementation

[0044] To enable those skilled in the art to better understand the technical solutions of this disclosure, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0045] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0046] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) involved in this disclosure require user authorization or full authorization from all parties when the embodiments of this disclosure are applied to specific products or technologies. Furthermore, the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions.

[0047] To facilitate understanding of the technical solutions provided in the embodiments of this application, some key terms used in the embodiments of this application will be explained below: Game Balance Testing (GBT) is an evaluation method for game design. It verifies the fairness of game characters, skills, economic systems, and other elements through methods such as simulating player behavior, data analysis, and adversarial experiments, ensuring that different strategies or gameplay styles have reasonable competitiveness.

[0048] Genetic Algorithm (GA) is an optimization algorithm that simulates the natural evolutionary process. It iteratively generates solutions through operations such as selection, crossover, and mutation, and is suitable for global search of complex problems and parameter optimization in machine learning.

[0049] Heuristic search (HS) is a search strategy based on experience or intuition. It guides the search direction through an evaluation function, significantly reducing computational load, and is often used in scenarios such as path planning and artificial intelligence decision-making.

[0050] Currently, the main technical solutions in the field of game balance testing include the following: (1) Manual testing method This method relies on the experience of game designers or testing teams, assessing balance by manually adjusting game parameters (such as character attributes, skill strength, resource output, etc.) and conducting simulated battles. Testers typically design specific battle scenarios, observe the performance of different strategies or character combinations, and make adjustments based on subjective judgment or simple data analysis. This method is usually optimized in conjunction with player feedback from the game's beta testing phase.

[0051] (2) A / B test A / B testing assesses game balance by dividing players into different groups, each experiencing different game parameter configurations (such as skill damage, economy system, etc.), and collecting real player behavior data (such as win rate, usage rate, game duration, etc.). This method relies on big data analysis to determine the optimal parameter configuration by comparing the performance of players in different groups. It is typically used for post-launch balance adjustments, such as hero strength optimization in MOBA games.

[0052] (3) Automated simulation testing Using rule-driven simulators, a large number of battle simulations are conducted in a virtual environment to statistically analyze indicators such as win rate and strength of different strategies or character combinations. For example, Monte Carlo simulation generates a large amount of battle data through random sampling to evaluate balance.

[0053] Manual testing methods are inefficient and limited in scale: Human testing is slow and struggles to cover a vast array of heroes, equipment, maps, and strategy combinations, severely limiting the breadth and depth of testing. They are also highly subjective and prone to bias: Test results heavily rely on the testers' personal experience, game understanding, and preferences, making it difficult to guarantee objectivity and comprehensiveness, and potentially overlooking unconventional playstyles or potential imbalances. Furthermore, they struggle to simulate real player behavior: The relatively fixed behavioral patterns of testing teams cannot fully simulate the diversity, creativity, and "irrational" strategies of the real player base, potentially failing to uncover balance issues that only surface when targeting a broad audience.

[0054] A / B testing carries the risk of trial and error, impacting player experience: Testing directly with real players can lead to some players experiencing a negative gaming experience if parameters are set incorrectly. It also has a long testing cycle and delayed feedback: Sufficient player data is required to draw conclusions, making it difficult to quickly validate adjustments and meet the demands of frequent, rapid balance changes. Furthermore, it is highly susceptible to external interference: Test results are easily influenced by player community opinions, popular gameplay styles ("version answers"), etc., and the data may not purely reflect the impact of parameter adjustments, resulting in high analytical complexity.

[0055] The AI ​​behavior in automated simulation tests differs from that of real players: Current AI intelligence and behavioral patterns differ significantly from those of human players, potentially causing test results to deviate from real-world scenarios and making it difficult to accurately predict players' strategies and reactions in actual matches. Simulation environments are complex and costly to build: Significant development resources are required to construct highly realistic game logic models and intelligent AI, resulting in high technical barriers and maintenance costs. They may fall into local optima or create "pseudo-equilibrium": Simulation tests may only achieve equilibrium within the AI's set strategy range, failing to capture innovative gameplay from human players. This could lead to "pseudo-equilibrium" combinations that only work in AI matches but are ineffective or overpowered in real-world matches.

[0056] To address the technical problems existing in current testing methods, this disclosure provides a business attribute updating method, apparatus, device, and storage medium. By fixing the front or back row attributes in the lineup, the search space is effectively limited to a few positions, significantly reducing the difficulty of searching for high-quality lineups, while promoting the generation of diverse solutions within a single row domain. On the one hand, it significantly reduces the search space; on the other hand, searching within a single row domain (front row lineup attribute information or back row lineup attribute information) is more likely to yield diverse suboptimal results, ensuring lineup diversity and reducing the risk of premature convergence. This increases the probability that the selected target lineup is a high-quality lineup and also improves the diversity of the selected multiple target lineups.

[0057] Please see Figure 1 The diagram illustrates an application environment for a business attribute update method according to an exemplary embodiment. The application environment may include server 01 and client 02.

[0058] Specifically, in this embodiment, server 01 may include a standalone server, a distributed server, or a server cluster composed of multiple servers. It may also be a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. Server 01 may include a network communication unit, a processor, and a memory, etc. Specifically, server 01 can be used to, according to the seed lineup pool, sequentially use the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup as the fixed attribute information of each seed lineup; then, fix the fixed attribute information of each seed lineup in the seed lineup pool, and iteratively update the non-fixed attribute information of each seed lineup to obtain an updated lineup pool, thereby determining the target lineup; and send the target lineup to client 02.

[0059] Specifically, in the embodiments of this specification, the client 02 may include physical devices such as smartphones, desktop computers, tablets, laptops, digital assistants, smart wearable devices, and in-vehicle terminals, and may also include software running on the physical device, such as web pages provided to users by some service providers, or applications provided to users by these service providers. Specifically, the client 02 can be used to display the target lineup.

[0060] Figure 2 This is a flowchart illustrating a business attribute update method according to an exemplary embodiment, such as... Figure 2 As shown, this method can be applied to Figure 1 The server 01 shown includes the following steps.

[0061] In step S201, the lineup attribute information set corresponding to each seed lineup in the seed lineup pool of the preset service is obtained; the lineup attribute information set includes at least two lineup attribute information.

[0062] In this embodiment, the preset business can be a game undergoing balance testing. The seed lineup pool can include at least two seed lineups, and the number of seed lineups in the seed lineup pool can be set according to actual conditions. The seed lineups can be initially constructed lineups. In the preset business scenario, the seed lineups can be the user's historical battle lineups in that scenario. Each seed lineup can include a set of lineup attribute information consisting of multiple lineup attribute information. For example, at the start of each main process, the top lineups produced in the historical main process are used as seeds (e.g., the maintenance scale is about 500), and evolutionary iterations are performed. Taking the game mechanism as an example of expanding from 3v3 to 5v5 and needing to consider the change in hero positioning order, the lineup combination space complexity increases exponentially, where n is the lineup attribute type.

[0063] Each lineup attribute can correspond to a lineup attribute type, which may include, but is not limited to, virtual character types and skill types in the game. For example, a lineup attribute type may include troop types and hero types in the game. Each lineup attribute type may include at least two lineup attribute pieces of information. For example, if the lineup attribute type is a troop type, its corresponding lineup attribute information may include archers, infantry, cavalry, etc. In different games, virtual character types may include different lineup attribute information. Each lineup (team) can contain several heroes, each with several skills. Heroes have different positions depending on the formation, and there is an implicit cyclical counter-relationship.

[0064] In step S203, for each seed lineup, the lineup attribute information set is divided into front-row lineup attribute information and back-row lineup attribute information according to the attribute type of each lineup attribute information, so as to obtain the front-row lineup attribute information and the back-row lineup attribute information corresponding to each seed lineup.

[0065] In this embodiment of the disclosure, the division between the front and back rows in the game is a core strategy for team composition, directly affecting the outcome of battles. The front and back rows in the team are the core parts of the tactical layout. The front row usually undertakes the responsibilities of tanking damage and controlling the enemy, while the back row is responsible for dealing damage and providing support.

[0066] Front-line units need high survivability, using positioning and equipment (such as tank items) to draw enemy fire and protect the back-line damage dealers. Front-line: Typically composed of tanks or high-survivability units, responsible for absorbing damage and protecting the backline; some special characters can temporarily act as front-line due to their mechanics. Back-line units need high damage output or control abilities, usually relying on the front-line for protection to avoid being killed by enemy assassins or burst damage. Back-line: Includes fragile units such as damage dealers (marksmen, mages) and supports (healers, controllers), who rely on the front-line for protection.

[0067] Each seed lineup can be divided into front-line and back-line units, which undergo alternating co-evolution during the main process—that is, optimizing the search only for the front-line or back-line each time. For example, the front-line typically consists of "tank" / "warrior" characters, while the back-line consists of "ranged marksman / mage" characters, with specific synergy relationships between characters of the same type. By fixing half of the lineup (front-line or back-line), the search space is effectively limited to 3 or 2 positions, significantly reducing the search difficulty while promoting diverse solutions within a single row. A row refers to a limited search space integrated with gameplay, such as the front / back area in a 5v5 battle. The specific definition of the row will be adjusted accordingly for different formations (square, circle, etc.) to ensure that the search strategy is closely integrated with the game mechanics. On the one hand, this significantly reduces the search space; on the other hand, searching within a single row makes it easier to generate diverse suboptimal results, ensuring lineup diversity and reducing the risk of premature convergence.

[0068] In step S205, the front row lineup attribute information or the back row lineup attribute information corresponding to each seed lineup is sequentially used as the fixed attribute information of each seed lineup; the fixed attribute information of each seed lineup in the seed lineup pool is fixed, and the non-fixed attribute information of each seed lineup is iteratively updated to obtain an updated lineup pool; the non-fixed attribute information of each seed lineup is the attribute information other than the fixed attribute information in the front row lineup attribute information and the back row lineup attribute information corresponding to each seed lineup.

[0069] In this embodiment, the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup can be used as the fixed attribute information of each seed lineup for iterative training. Then, the fixed attribute information of each seed lineup in the seed lineup pool is fixed, and the non-fixed attribute information of each seed lineup is iteratively updated to obtain an updated lineup pool. For example, the front-row lineup attribute information corresponding to each seed lineup can be used as the fixed attribute information for one update, and then the back-row lineup attribute information in the updated lineup can be used as the fixed attribute information of each seed lineup for another iterative update. Thus, an updated lineup pool including multiple updated lineups can be obtained.

[0070] In step S207, the target lineup is determined based on the battle results between lineups in the updated lineup pool.

[0071] In this embodiment of the disclosure, one or more target lineups can be selected from the battle results between the lineups in the updated lineup pool, and lineups can be recommended to users corresponding to preset services.

[0072] This embodiment of the disclosure obtains a set of lineup attribute information corresponding to each seed lineup in a seed lineup pool of a preset service; the set of lineup attribute information includes at least two sets of lineup attribute information; for each set of lineup attribute information corresponding to a seed lineup, the set of lineup attribute information is divided into front-row lineup attribute information and back-row lineup attribute information according to the attribute type corresponding to each set of lineup attribute information, to obtain the front-row lineup attribute information and the back-row lineup attribute information corresponding to each seed lineup; then, the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup is sequentially used as the fixed attribute information of each seed lineup; the fixed attributes of each seed lineup in the seed lineup pool are fixed. The system collects information and iteratively updates the non-fixed attribute information of each seed lineup to obtain an updated lineup pool. The non-fixed attribute information of each seed lineup includes the attribute information of the front-row lineup and the attribute information of the back-row lineup, excluding the fixed attribute information. By fixing the front-row or back-row attributes in the lineup, the search space is effectively limited to a few positions, significantly reducing the difficulty of searching for high-quality lineups and promoting the generation of diverse solutions within a single row domain. On the one hand, it significantly reduces the search space; on the other hand, searching within a single row domain (front-row lineup attribute information or back-row lineup attribute information) is more likely to produce diverse suboptimal results, ensuring lineup diversity and reducing the risk of premature convergence. Finally, based on the battle results between lineups in the updated lineup pool, the target lineup is determined. This can increase the probability that the selected target lineup is a high-quality lineup and increase the diversity of the selected target lineups.

[0073] In some embodiments, such as Figure 3 As shown, the step of sequentially using the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup as the fixed attribute information of each seed lineup; fixing the fixed attribute information of each seed lineup in the seed lineup pool, and iteratively updating the non-fixed attribute information of each seed lineup to obtain an updated lineup pool includes: S2051: The front row lineup attribute information or the back row lineup attribute information corresponding to each seed lineup is taken as the first lineup attribute information of each seed lineup; and the attribute information other than the first lineup attribute information in the front row lineup attribute information and the back row lineup attribute information corresponding to each seed lineup is determined as the second lineup attribute information of each seed lineup. S2053: Fix the second lineup attribute information of each seed lineup in the seed lineup pool, and iteratively update the first lineup attribute information of each seed lineup to obtain an initial updated lineup pool; the initial updated lineup pool includes the second lineup attribute information and the first updated lineup attribute information corresponding to each initial updated lineup; the first updated lineup attribute information is the attribute information after the first lineup attribute information is updated. S2055: Based on the battle results between the initial updated lineup in the initial updated lineup pool and the seed lineup in the seed lineup pool, construct the updated lineup pool.

[0074] In this embodiment, the second lineup attribute information can be used as fixed attribute information first; then, the first lineup attribute information can be used as fixed attribute information for iterative lineup updates. For example, the front-row lineup attribute information corresponding to each seed lineup can be used as the first lineup attribute information for each seed lineup, and the back-row lineup attribute information corresponding to each seed lineup can be used as the second lineup attribute information for each seed lineup. In other embodiments, the back-row lineup attribute information corresponding to each seed lineup can be used as the first lineup attribute information for each seed lineup, and the front-row lineup attribute information corresponding to each seed lineup can be used as the second lineup attribute information for each seed lineup. Then, the second lineup attribute information of each seed lineup is fixed, and the first lineup attribute information of each seed lineup is iteratively updated to obtain an initial updated lineup pool. Each initial lineup in the initial updated lineup pool includes fixed second lineup attribute information and first updated lineup attribute information obtained after updating based on the first lineup attribute information. Afterwards, the battle results between the initial updated lineups in the initial updated lineup pool and the seed lineups in the seed lineup pool can be compared to construct an updated lineup pool that includes high-quality lineups. This ensures that the overall performance of the updated lineup pool is better than that of the seed lineup pool, so that lineup optimization can be performed based on the seed lineup pool to improve the lineup quality.

[0075] In some embodiments, constructing the updated lineup pool based on the battle results between the initial updated lineups in the initial updated lineup pool and the seed lineups in the seed lineup pool includes: Obtain the parent-child team battle results between each initial updated team in the initial updated team pool and each seed team in the seed team pool; Based on the battle results, a selection pool is determined from the initial updated lineup pool and the seed lineup pool; The first lineup attribute information of each lineup in the lineup pool is fixed, and the second lineup attribute information of each lineup is iteratively updated to obtain the second updated lineup attribute information. Based on the first lineup attribute information and the second updated lineup attribute information corresponding to each selected lineup, an updated lineup is constructed; and based on the updated lineup, the updated lineup pool is constructed.

[0076] In this embodiment, since the initial updated lineup pool is formed based on the seed lineup pool, the seed lineup pool can be considered as the parent pool and the initial updated lineup pool as the child pool. The selection lineup pool can be determined based on the overall battle results between the parent and child pools. For example, the selection lineup pool can be determined from the two pools based on the overall battle results between the initial updated lineup pool and the seed lineup pool. If the overall battle results of the initial updated lineup pool are better than those of the selection lineup pool, the initial updated lineup pool can be used as the seed lineup pool for the next iteration. If the overall battle results of the seed lineup pool are better than those of the selection lineup pool, the current lineup pool can continue to be used as the seed lineup pool for the next iteration. The seed lineup pool serves as the seed lineup pool for the next iteration, i.e., the selection lineup pool. Then, in the next iteration, the lineup attribute information in the selection lineup pool that was not updated in the previous round can be updated. For example, the first lineup attribute information of each selection lineup in the selection lineup pool can be fixed, and the second lineup attribute information of each selection lineup can be iteratively updated to obtain the second updated lineup attribute information. Based on the first lineup attribute information and the second updated lineup attribute information corresponding to each selection lineup, the lineup attribute information in the lineup can be classified and updated in batches to obtain updated lineups. And based on the updated lineups, the updated lineup pool is constructed.

[0077] This embodiment can effectively limit the search space to a few positions by fixing the front or back row attributes in the lineup, significantly reducing the difficulty of searching for high-quality lineups, while promoting the generation of diverse solutions within a single row domain; on the one hand, it significantly reduces the search space; on the other hand, searching within a single row domain (front row lineup attribute information or back row lineup attribute information) is more likely to produce diverse suboptimal results, ensuring lineup diversity and reducing the risk of premature convergence.

[0078] In some embodiments, the advantageous lineups in the initial updated lineup pool and the seed lineup pool can be selected based on the combined battle results between the parent and child lineups of the same lineup, and a selection lineup pool can be constructed. Specifically, each seed lineup in the seed lineup pool can be used as a parent lineup, and the corresponding seed lineup in the initial updated lineup pool can be used as a child lineup. Then, the selection lineup is determined based on the battle results between the parent and child lineups. For example, if the selection lineup is the initial updated lineup, the first lineup attribute information of the selection lineup is the first updated lineup attribute information; if the selection lineup is the seed lineup, the first lineup attribute information of the selection lineup is the first lineup attribute information of the seed lineup.

[0079] In some embodiments, fixing the second lineup attribute information of each seed lineup in the seed lineup pool and iteratively updating the first lineup attribute information of each seed lineup to obtain an initial updated lineup pool includes: The second lineup attribute information of each seed lineup in the seed lineup pool is fixed, and a lineup cross operation is performed based on the first lineup attribute information of each seed lineup to obtain the first updated lineup; Based on the first lineup attribute information of each seed lineup, a lineup mutation operation is performed to obtain a second updated lineup; The initial updated lineup pool is constructed based on the first updated lineup and the second updated lineup.

[0080] In this embodiment of the disclosure, during the iterative update of the lineup, lineup updates can be performed through operations such as lineup crossover and lineup mutation. For example, the second lineup attribute information of each seed lineup in the seed lineup pool can be fixed, and a lineup crossover operation can be performed based on the first lineup attribute information of each seed lineup to obtain the first updated lineup. In the lineup crossover stage, two lineups from the seed lineup pool are randomly sampled based on the current population: lineup a and lineup b (only the parent generation is sampled; the new offspring in the current round do not participate in the crossover). A position i is randomly selected within the row domain of lineup a, and the entire "hero + its carried skills and fate" at position j in lineup b is replaced to the i-th position in lineup a to obtain a new lineup. If "hero duplication, skill duplication, fate incompatibility, illegal formation" or duplication with an existing population occurs, the lineup is discarded and retried a limited number of times. A fixed number of new lineups (e.g., 50) are generated each round, and it is possible to select only lineups that have not appeared before. Then, based on the first lineup attribute information of each seed lineup, a lineup mutation operation is performed to obtain the second updated lineup. In the lineup mutation stage, the parent generation is used as the benchmark, and multiple mutation types are triggered according to probability and restricted to the current row domain to randomly generate a certain number of new lineups (e.g., 100).

[0081] For example, lineup mutation specifically includes five methods: Formation mutation: replacing with a different formation and refreshing the team order through equivalent point normalization. Hero mutation: randomly replacing several positions within the row with new heroes that are not currently occupied (retaining the original skills). Skill mutation: randomly replacing several positions within the row with new skills that are not currently occupied. Hero order mutation: performing several pairwise swaps within the row (limited to the same row). Fate mutation: re-collecting fates for several heroes within the row (excluding the current item from the hero's available fate set).

[0082] Finally, the first and second updated lineups were determined as the initial updated lineups, thereby increasing the diversity of lineups in the initial updated lineup pool.

[0083] This embodiment establishes a highly flexible lineup data structure, which not only includes basic dimensions such as hero configuration, skill combination, formation layout, positioning strategy, and destiny attributes, but more importantly, it supports dynamic expansion. It can flexibly add or remove diverse game elements such as equipment system, horse mounts, treasure items, runes and inscriptions according to specific game needs, ensuring that the solution can adapt to game types with different complexities and mechanics.

[0084] In some embodiments, obtaining the parent-child team battle results between each initial updated team in the initial updated team pool and each seed team in the seed team pool includes: Obtain each initial updated lineup from the initial updated lineup pool as the parent lineup; and obtain the seed lineup from the seed lineup pool as the child lineup. The battle result of the parent and child lineups is determined based on their win rate, survival rate, and loss ratio in a preset business scenario.

[0085] In this embodiment, each initial updated lineup in the initial updated lineup pool can be obtained as a parent lineup; and a seed lineup in the seed lineup pool can be obtained as a child lineup. Then, the win rate, survival rate, and kill ratio of the parent and child lineups under a preset business scenario are obtained. The preset business scenario refers to one or more scenarios under a preset business. For example, in a game scenario, the preset business scenario may include, but is not limited to, a round-robin scenario or a tournament scenario. A round-robin is a competition method where participants compete against all other opponents one or more times, and their ranking is determined based on data such as points, win-loss relationship, or point differential. Its main forms include single round-robin, double round-robin, and group round-robin. Round-robin scenarios can include full round-robin and non-full round-robin. Then, based on the win rate, survival rate, and kill ratio of the parent and child lineups under the preset business scenario, the parent-child lineup battle results can be determined based on multiple indicators, improving the accuracy of the battle results and thus increasing the probability of selecting high-quality lineups from the lineup pool.

[0086] In some embodiments, determining the battle result of the parent and child lineups based on their win rate, survival rate, and casualty ratio in a preset business scenario includes: Obtain the win rate, survival rate, and kill ratio of the parent lineup and the child lineup in the round-robin scenario to obtain the first match result; The win rate, survival rate and casualty ratio of the parent lineup and the child lineup in the tournament scenario are obtained to obtain the second match result. The results of the father-son team battle are determined based on the results of the first and second battles.

[0087] In this embodiment of the disclosure, the preset business scenario may include a round-robin scenario and a tournament scenario; the win rate, survival rate, and kill ratio of the parent lineup and the child lineup in the round-robin scenario can be obtained to obtain the first match result; for example, the round-robin scenario may include a full round-robin and a non-full round-robin, and the win rate, survival rate, and kill ratio of the parent lineup and the child lineup in the full round-robin scenario, as well as the win rate, survival rate, and kill ratio in the non-full round-robin scenario, can be obtained sequentially; then, based on the win rate, survival rate, and kill ratio in the full round-robin scenario, and the win rate, survival rate, and kill ratio in the non-full round-robin scenario, the first match result is determined.

[0088] For example, weights can be set for win rate, survival rate, and kill ratio respectively. Then, based on the weights of win rate, survival rate, and kill ratio in the full-round-robin scenario, the full-round battle results for the parent and child lineups in the full-round-robin scenario can be determined. Based on the weights of win rate, survival rate, and kill ratio in the non-full-round-robin scenario, the non-full-round battle results for the parent and child lineups in the non-full-round-robin scenario can be determined. Furthermore, weights of indicators can be set for the full-round-robin and non-full-round-robin scenarios, and a weighted sum can be performed based on the full-round and non-full-round battle results of the lineups, as well as the weights of indicators in the full-round and non-full-round-robin scenarios, to obtain the parent-child lineup battle results. This allows the battle results to be determined based on indicator data of two lineups in multiple scenarios, improving the accuracy of battle result evaluation and thus improving the accuracy of selecting high-quality lineups.

[0089] In some embodiments, within a game scenario, core performance metrics (such as win rate, survival rate, and kill ratio) that the game designers focus on can be set as evaluation metrics; these metrics serve as the comprehensive optimization objective function during the genetic search process. For example, win rate and kill ratio can be selected as evaluation metrics, and the objective function can be calculated as: 0.6 * win rate + 0.4 * kill ratio; where 0.6 is the weight of win rate and 0.4 is the weight of kill ratio, thereby driving the strategy combination to continuously approach the near-optimal solution link based on the objective function.

[0090] In some embodiments, such as Figure 4 As shown, determining the selected lineup from the initial updated lineup pool and the seed lineup pool based on the battle results includes: S401: Based on the lineup attribute information corresponding to the parent lineups in the seed lineup pool, determine the character similarity, skill similarity, and formation similarity between the parent lineups; S403: Determine the lineup difference results of the seed lineup pool based on the character similarity, skill similarity, and formation similarity among the parent lineups; S405: Based on the lineup attribute information corresponding to the offspring lineups in the initial updated lineup pool, determine the character similarity, skill similarity, and formation similarity among the offspring lineups; S407: Determine the lineup difference results of the initial updated lineup pool based on the character similarity, skill similarity, and formation similarity among the offspring lineups; S409: Based on the battle results, the lineup difference results of the seed lineup pool, and the lineup difference results of the initial updated lineup pool, a selection lineup is determined from the initial updated lineup pool and the seed lineup pool.

[0091] In this embodiment of the disclosure, the lineup attribute information may include, but is not limited to, information such as characters, skills, and formations. During the lineup iteration and update process, the character similarity, skill similarity, and formation similarity between lineups in the same lineup pool can also be calculated to determine the lineup difference results in the lineup pool. Then, based on the battle results between the two generations of lineups, the lineup difference results of the seed lineup pool, and the lineup difference results of the initial updated lineup pool, the selected lineup can be determined from the initial updated lineup pool and the seed lineup pool.

[0092] In the game scenario, character similarity can be hero similarity. This embodiment quantifies the differences in multiple dimensions (such as hero similarity, skill similarity, and formation similarity) and assigns differentiated importance weights to the similarity of different combinations. This can prevent the algorithm from converging to a local optimum trap too early and ensure that the test results cover a sufficiently broad strategy space, effectively avoiding the omission of potential balance problems and hidden strong combinations.

[0093] In this embodiment, the evaluation index system includes both core strength indicators (such as win rate, survival rate, and kill ratio) that game designers care about, and diverse indicators that reflect test coverage and strategy richness, to fully reflect the true performance of different strategy gameplay. The former serves as a comprehensive optimization objective function in the genetic search process, driving strategy combinations to continuously approach the near-optimal solution path; the latter plays a key role in maintaining population diversity and regulating selection pressure during genetic mutation. By quantifying multi-dimensional (such as hero similarity, skill similarity, and formation similarity) difference indicators and assigning differentiated importance weights to the similarity of different combinations, it can prevent the algorithm from prematurely converging to local optima and ensure that the test results cover a sufficiently broad strategy space, effectively avoiding the omission of potential balance issues and hidden strong combinations.

[0094] In some embodiments, determining the selected lineups from the initial updated lineup pool and the seed lineup pool based on the battle results, the lineup difference results of the seed lineup pool, and the lineup difference results of the initial updated lineup pool includes: Based on the battle results, the first battle score of the seed lineup pool and the second battle score of the initial updated lineup pool are determined; Based on the first battle score and the lineup difference results of the seed lineup pool, the comprehensive evaluation score of the seed lineup pool is determined; Based on the second battle score and the lineup difference results of the initial updated lineup pool, the comprehensive evaluation score of the initial updated lineup pool is determined; Based on the comprehensive evaluation score of the seed lineup pool and the comprehensive evaluation score of the initial updated lineup pool, a selection lineup is determined from the initial updated lineup pool and the seed lineup pool.

[0095] In this embodiment of the disclosure, determining the selected lineup from the initial updated lineup pool and the seed lineup pool may specifically include: determining a first battle score for the initial updated lineup pool and a second battle score for the seed lineup pool based on the battle results of the two lineup pools; thereby quantifying the battle results and improving the accuracy of lineup performance evaluation; for example, a higher battle score can be set for the lineup pool with higher performance index data, and a lower battle score can be set for the lineup pool with lower performance index data; and the difference evaluation score of the two lineup pools can be determined based on the lineup difference results corresponding to each of the two lineup pools, thereby obtaining the first evaluation score of the initial updated lineup pool and the second evaluation score of the seed lineup pool; for example, the sum of the first battle score and the first evaluation score can be calculated to obtain the seed lineup. The comprehensive evaluation score of the seed lineup pool is calculated; the sum of the second battle score and the second evaluation score is calculated to obtain the comprehensive evaluation score of the initial updated lineup pool; in some embodiments, a first preset weight corresponding to the battle score and a second preset weight corresponding to the difference index can be set, and then the first product of the first battle score and the first preset weight and the second product of the first evaluation score and the second preset weight are calculated respectively; the sum of the first product and the second product is calculated to obtain the comprehensive evaluation score of the seed lineup pool; the third product of the second battle score and the first preset weight and the fourth product of the second evaluation score and the second preset weight are calculated respectively; the sum of the third product and the fourth product is calculated to obtain the comprehensive evaluation score of the initial updated lineup pool; then the lineups in the seed lineup pool and the lineup pool with higher comprehensive evaluation scores in the initial updated lineup pool can be determined as the screening lineups.

[0096] This embodiment determines the comprehensive evaluation score of the seed lineup pool based on the first battle score and the lineup difference results of the seed lineup pool; it determines the comprehensive evaluation score of the initial updated lineup pool based on the second battle score and the lineup difference results of the initial updated lineup pool; and then, based on the comprehensive evaluation scores of the seed lineup pool and the initial updated lineup pool, it ensures the overall performance of the selected lineup.

[0097] In some embodiments, determining the lineup difference results of the seed lineup pool based on the character similarity, skill similarity, and formation similarity between the parent lineups includes: The first difference result is determined based on the role similarity between the parent generation lineups; The second difference result is determined based on the skill similarity between the parent lineups; The third difference result is determined based on the formation similarity between the parent lineups; Obtain the first weight corresponding to the character similarity, the second weight corresponding to the skill similarity, and the third weight corresponding to the formation similarity; The lineup difference result of the seed lineup pool is determined based on the product of the first difference result and the first weight, the product of the second difference result and the second weight, and the product of the third difference result and the third weight.

[0098] In this embodiment, the character similarity, skill similarity, and formation similarity between two lineups in the parent lineup pool can be calculated sequentially. Weights can be set for each of the character similarity, skill similarity, and formation similarity, allowing for a weighted summation to obtain the lineup difference result. For example, the product of each difference result and its corresponding weight can be calculated first, and then the sum of these products can be obtained to obtain the lineup difference result of the parent lineup pool. In some embodiments, a similar method can be used to determine the lineup difference result of the initial updated lineup pool; thus, a comprehensive evaluation of the lineup difference results in the lineup pool can be achieved from multiple dimensions such as characters, skills, and formations.

[0099] In some embodiments, constructing the updated lineup pool based on the battle results between the initial updated lineups in the initial updated lineup pool and the seed lineups in the seed lineup pool includes: Obtain the battle results between the various initial updated lineups in the initial updated lineup pool; Based on the battle results between the initial updated lineups, candidate lineups are selected from the initial updated lineup pool to construct a candidate lineup pool; Based on the battle results between the candidate lineup pool and the seed lineups in the seed lineup pool, the updated lineup pool is constructed using a restricted tournament replacement strategy.

[0100] In this embodiment, the battle results between the initial updated lineups in the initial updated lineup pool can be obtained. Then, based on the battle results between the initial updated lineups, candidate lineups are selected from the initial updated lineup pool to construct a candidate lineup pool. For example, the lineups in the initial updated lineup pool can be sorted from largest to smallest according to the battle result score, and the top k lineups can be selected as candidate lineups based on the sorting result. Alternatively, a score threshold can be set, and lineups with battle result scores greater than the preset threshold can be selected as candidate lineups, thereby constructing a candidate lineup pool. Then, based on the battle results between the candidate lineup pool and the seed lineups in the seed lineup pool, the Restricted Tournament Selection (RTS) algorithm is used to construct the updated lineup pool. RTS is an improved selection strategy in genetic algorithms that optimizes population diversity or accelerates convergence by limiting the range of individuals participating in the tournament. The following are its core points: Restricted range: Participants are selected only from the local neighborhood (such as a circular topology or niche) of the current individual, rather than the global population. Reduced competitive pressure: Reduces direct competition between individuals in the population and avoids premature elimination of potentially high-quality solutions. Increased diversity: Local selection reduces premature monopolization of the global optimum, preserving more potential solutions. Adaptation to multimodal problems: In multimodal optimization, local tournaments can locate multiple local optima.

[0101] In this embodiment, a steady-state replacement is attempted using the RTS algorithm to maintain a constant population size. For each offspring, a window w (e.g., w = 50) is randomly selected from the parents to calculate "team similarity." If there are ties among the most similar parents, the one with the lowest fitness is selected as the competitor. Replacement is only performed if the offspring's fitness is higher than that of the parent; otherwise, the offspring is discarded. The similarity can be calculated using a linear weighted average of "skill overlap × 0.3 + hero overlap × 0.8 + formation similarity × 0.7". At the end of each round, only the scores of the surviving individuals are retained; the historical scores of eliminated individuals are not used to avoid creating a lingering bias in subsequent rounds.

[0102] For example, the parent lineup pool corresponding to each child lineup pool can be obtained, and w lineups can be randomly selected from the parent lineup pool according to a preset window number to form multiple parent lineup combinations. The similarity between each parent lineup combination and the child lineup pool can be calculated, thereby selecting the lineup combination with the highest similarity to the child lineup pool from the multiple parent lineup combinations as the target parent combination. If there are multiple target parent combinations, the fitness corresponding to each target parent combination can be calculated, and the target parent combination with the lowest fitness can be selected as the filter parent combination. Then, the fitness of the filter parent combination is compared with the fitness of the child lineup pool. If the fitness of the child lineup pool is higher than the fitness of the filter parent combination, the child lineup pool is used to replace the filter parent combination as the seed lineup for the next round of iteration update.

[0103] In some embodiments, determining the target lineup based on the battle results between lineups in the updated lineup pool includes: Based on the updated lineup pool, construct at least two lineup combinations; Obtain the battle results for each lineup combination in at least two preset business scenarios; Based on the battle results of each lineup combination in at least two preset business scenarios, the target lineup is selected from the updated lineup pool; The method further includes: The system sends the lineup attribute information of the target lineup to the terminal so that the terminal can display the lineup attribute information of the target lineup.

[0104] In this embodiment of the disclosure, during the lineup iteration update process, the update can be terminated after the target number of updates, or after the lineups in the updated lineup pool meet preset conditions. After the lineup iteration update is completed, the lineups in the updated lineup pool can be randomly combined according to a preset number to construct at least two lineup combinations. Then, the battle results corresponding to the lineups in each lineup combination under at least two preset business scenarios are obtained. The at least two preset business scenarios may include, but are not limited to, round-robin scenarios and tournament scenarios. Based on the battle results of the lineup combinations under multiple preset business scenarios, high-quality target lineups can be selected.

[0105] In one exemplary embodiment, such as Figure 5 As shown, Figure 5 This diagram illustrates the match lineups in full round-robin, non-full round-robin, and tournament scenarios. In the full round-robin scenario, N represents the total number of lineups in the lineup pool. All lineups (N) are taken from the lineup space and used in a round-robin format. The total number of matches is: In this scenario, the win rate and storage rate of each lineup are obtained as true values ​​of strength, with the highest accuracy, but the longest time consumption, requiring a number of battles.

[0106] In a non-full round-robin scenario, K is the number of lineups, and K << N. The K lineups are compared in a cycle, and the number of matches required is: In this scenario, the obtained win rate and storage rate are approximate estimates of the true strength value for a specific group. The higher the overall strength of the group, the closer it is to the true value, and the number of matches required is.

[0107] In a tournament scenario, K1 is the number of challenging lineups, and K2 is the number of benchmark lineups. Among them, K1 >> K2, and the strength of the benchmark lineups is relatively high; each challenging lineup plays against each benchmark lineup m times, and the number of matches required is: Taking the full round-robin strength as the true value, from left to right, the accuracy of strength estimation becomes lower and lower; but the number of matches decreases, and the time cost becomes lower. Different evaluation scenarios can be adopted in different links.

[0108] In this scenario, the win rate and storage rate are obtained under a specific benchmark lineup (seed lineup), which is an approximate estimate of the true strength value. The higher the strength of the benchmark lineup, the closer it is to the true value, and the number of matches required is.

[0109] Exemplarily, planners usually hope to find the "strongest lineup", but the definition and verification of "strongest" are controversial both in theory and practice. Due to complex game relationships such as unit countering, almost every lineup can be countered by another targeted lineup, and this non-transitivity makes it difficult to define absolute strength. The countering relationships can be divided into two categories: explicit countering (such as cavalry countering infantry) has a clear numerical bonus and a significant impact, and can be optimized separately through unit classification; while implicit countering (such as assassins countering ranged units) stems from tactical mechanism interactions, which are more complex and difficult to simply classify, and are also the core balance issues that planners focus on. Therefore, this solution combines two evaluation methods, "round-robin" and "tournament", to balance the calculation cost while ensuring accuracy, and improves the reliability of the results through multi-dimensional verification.

[0110] As Figure 6 shown, Figure 6 is a flowchart of a lineup update method in multiple scenarios, which includes multiple loops of the main process for lineup update. Each main process runs through the genetic evolution search stage, strength update stage, and visualization result stage in a complete manner; The team has a formation attribute. Each team contains several heroes. The heroes carry several skills, and the heroes have different positions according to different formations. Taking the example of an invisible cycle of countering scenarios, this solution is illustrated; this solution also supports flexible expansion of the numerical system, such as equipment, mounts, etc.

[0111] 1. Genetic search stage At the start of each main process, the top lineups from previous main processes are used as seeds (e.g., with a maintenance scale of approximately 500) for iterative evolution. Taking the game mechanics expanding from 3v3 to 5v5, and the need to consider changes in hero positioning order, as an example, the complexity of lineup combination space increases exponentially. To address this challenge, this invention, considering the characteristics of the game mechanics, divides each seed lineup into front-row and back-row components, alternating them for co-evolution during the main process—that is, each time only the front-row or back-row is optimized and searched.

[0112] The front row typically consists of "tank" / "warrior" type characters, while the back row consists of "ranged archer / mage" type characters, with specific synergy relationships between characters of the same type. By fixing half of the lineup (front row or back row), the search space is effectively limited to 3 or 2 positions, significantly reducing the difficulty of searching while promoting the generation of diverse solutions within a single row. A row refers to a limited search space combined with gameplay, such as the front / back row area in a 5v5 battle. The specific definition of the row will be adjusted accordingly for different formations (square formation, circle formation, etc.) to ensure that the search strategy is closely integrated with the game mechanics. On the one hand, this significantly reduces the search space; on the other hand, searching within a single row makes it easier to generate diverse suboptimal results, ensuring lineup diversity and reducing the risk of premature convergence.

[0113] Starting from the base population, iterate through several genetic rounds (e.g., 150 rounds), with each round following the sequence of "crossover → mutation → evaluation → survival of the fittest". The termination condition can be configured to stop at a fixed number of rounds or when an improvement threshold is reached; after termination, switch to another row for co-evolution, forming a gradual optimization with alternating "front / back rows".

[0114] During the iteration process, the stages of "roster crossover", "roster mutation", "roster evaluation" and "survival of the fittest" will be carried out in sequence.

[0115] In the crossover phase, two teams, a and b, are randomly sampled from the current population (only the parent generation is sampled; new offspring in the current round are not included in the crossover). Within the row domain of team a, a position i is randomly selected. The entire "hero + their skills and destiny" of any position j in team b is replaced with the result of team a[i], resulting in a new individual. If a new individual is found due to "duplicate hero, duplicate skills, incompatible destiny, illegal formation," or duplication with an existing population, it is discarded and the process is retried a limited number of times. A fixed number of new individuals are generated each round (e.g., 50), and only valid and unseen results are retained.

[0116] In the lineup mutation phase, based solely on the parent generation, multiple mutation types are triggered with probability and limited to the current row domain, randomly generating a certain number of new lineups (e.g., 100). The mutations specifically include 5 types: Formation Variation: Replace with different formations and refresh the team order through equal point normalization.

[0117] Hero Mutation: Randomly replaces several unoccupied positions within the row with new heroes (retaining original skills).

[0118] Skill Mutation: Randomly replaces several positions within the row with new, unoccupied skills.

[0119] Hero order mutation: Performs several pairwise swaps within the row (only within the same row).

[0120] Fate Mutation: Re-harvest Fates for several heroes within the domain (excluding the current item from the hero's available Fate set).

[0121] In the lineup evaluation phase, a challenge simulation is conducted for "parent generation ∪ offspring generation". Fitness is calculated using a comprehensive index (e.g., score = 0.6 × win rate + 0.4 × survival rate). Each opponent undergoes multiple independent simulations, and the average score is taken (e.g., 3 games / opponent). Individual-level caching is employed: historical scores of previously encountered teams are directly reused, and matches are only played against newly encountered individuals for the first time. Parallel evaluation is performed using a process pool, and a baseline opponent set is set once during the process initialization phase to reduce overhead.

[0122] The survival-of-the-fittest process employs the RTS algorithm: attempting steady-state replacements to maintain a constant population size. For each offspring, a window w (e.g., 50) is randomly selected from the parents to calculate "team similarity." If there are ties among the most similar parents, the one with the lowest fitness is selected as the competitor. Replacement is only performed if the offspring's fitness is higher than that of the parent; otherwise, the offspring is discarded. Similarity is calculated using a linear weighted average of "skill overlap × 0.3 + hero overlap × 0.8 + formation similarity × 0.7". At the end of each round, only the scores of surviving individuals are retained; the historical scores of eliminated individuals are not used to avoid bias in subsequent rounds.

[0123] During the strength update phase, such as Figures 7-8 As shown, Figure 7 This displays a visualization of the top teams from the round-robin tournament after a complete main round (including multiple iterations) has concluded, after deduplication by aggregating similar teams. Figure 8 This demonstrates the results when merging search results from multiple main loops (T+x rounds). The correlation between the outputs of each round is weak, revealing two common issues overall: Local Optimality Dilemma: Because some general combinations have additional benefits such as allusions / sets / bonds, cross-pollination or mutation around such templates often loses the benefits, resulting in new individuals generally becoming weaker. The search is easily "stuck" near the local extreme value, making it difficult to leap forward.

[0124] Environment-dependent bias: "Environment" refers to the baseline opponent set used in the evaluation. Even if the influence of cyclical counters is weakened by aligning the unit proportions, bias is still difficult to avoid: if the baseline is more focused on burst damage, resilient / durable lineups are more likely to have an advantage in confrontations with the baseline; if the baseline is more focused on sustain / recovery, assassination / burst damage lineups are more likely to be evaluated as relatively stronger.

[0125] To address the aforementioned issues and to more comprehensively demonstrate the strength of different gameplay strategies and fully showcase their balance, this solution also employs a baseline lineup strength update and RTS replacement algorithm. The baseline lineup strength is updated by performing a round-robin comprehensive evaluation with the parent population after each main process genetic search. This comprehensive evaluation avoids testing errors caused by insufficient evaluation dimensions in tournaments. Based on the comprehensive score of this round-robin, the baseline set is continuously supplemented or replaced from the candidate tops, while ensuring novelty (no duplicate character activation codes) and diversity (formation / style coverage), maintaining a fixed size to avoid "one round determining the entire team." Additional challenges and updates are generated by randomly sampling from the parent population and taking half from the current round's top candidates for each main process to create a new additional challenge set. This random sampling provides greater robustness against adversaries, while the sampling of top lineups ensures the real-time strength of the baseline lineup as the genetic search progresses.

[0126] Seed pool update, such as Figure 9 As shown, Figure 9 This diagram illustrates a seed pool update method. It involves obtaining the parent lineup pool corresponding to each child lineup pool, and randomly selecting w lineups from the parent lineup pools according to a preset window size to form multiple parent lineup combinations. The similarity between each parent lineup combination and the child lineup pool is calculated, thus selecting the lineup combination with the highest similarity to the child lineup pool as the target parent combination. If multiple target parent combinations exist, the fitness of each target parent combination is calculated, and the target parent combination with the lowest fitness is selected as the filter parent combination. Then, the fitness of the filter parent combination is compared with the fitness of the child lineup pool. If the fitness of the child lineup pool is higher than that of the filter parent combination, the child lineup pool replaces the filter parent combination as the seed lineup for the next iteration. The "parent (previous round starting pool)" and "current round Top candidates" are selectively retained using Restricted Tournament Replacement (RTS), competing based on window nearest neighbor similarity and fitness, with the size remaining unchanged, serving as the starting seed pool for the next main process.

[0127] The visualization phase aims to present the results of the genetic search test from different dimensions, providing an assessment of its strength and showcasing the balance under different gameplay strategies. All charts support interactive conditional filtering. The tournament's strength for each formation is summarized and displayed based on the results of challenge matches against a benchmark strong set (including additional challenge opponents). The strength distribution and tiers are presented by formation type. Viewers can see the strength ranking and gaps of each formation in this round, providing a direct comparison of the strength position of the "newly generated formations" relative to the benchmark set. The strength ranges and tiers are summarized by formation type based on the results of matches against the benchmark strong set (including additional challenge opponents).

[0128] The round-robin tournament deduplicated the strength of each formation, aggregated similar lineups to remove duplicates, and retained only the strongest representative samples. The strength distribution was displayed by formation to avoid interference from similar results and to more clearly reflect the strength ceiling and stability zone of each formation. The number of times each hero is selected in the front and back rows is recorded, and the appearance and coverage of heroes in the front and back rows are statistically analyzed to identify key heroes and provide an intuitive reference for lineup and numerical adjustments.

[0129] In this embodiment, a multi-dimensional evaluation index framework covering the core concerns of game planning and operation is constructed, including intuitive performance indicators such as win rate, survival rate, and kill ratio, as well as comprehensive balance indicators such as adaptability and diversity, to ensure that the game balance and intensity distribution under different strategy gameplay can be fully reflected.

[0130] Evaluation Scenarios: To effectively balance testing accuracy and computational resource consumption, a multi-level evaluation scenario system was designed. With win rate as the core, differentiated testing modes were established, including full round-robin (single lineup against all possible lineups), round-robin (single lineup against a specific group), and tournament (single lineup against some strong lineups), achieving a dynamic balance between accuracy and efficiency.

[0131] Combat Simulator: Employing a highly modular architecture, it completely decouples combat logic, rule engine, and balance testing algorithms, supporting rapid integration of combat mechanics from different game genres. The simulator boasts powerful parallel computing capabilities, enabling efficient execution of massive amounts of combat simulations. Furthermore, its standardized interface design ensures rapid migration and customized adaptation across different game projects.

[0132] Team Composition: A highly flexible team composition data structure has been established, which not only includes basic dimensions such as hero configuration, skill combinations, formation layout, positioning strategy, and destiny attributes, but more importantly, it supports dynamic expansion. It can flexibly add or remove diverse game elements such as equipment systems, horse mounts, treasure items, runes and inscriptions according to specific game needs, ensuring that the solution can adapt to game types with different complexities and mechanics.

[0133] Main Process: Considering that the game will undergo flexible numerical adjustments and mechanism optimizations based on balance test results during continuous iteration, this solution adopts a continuous running mode, with the main process executing cyclically, each round taking approximately one day. By inheriting only the optimization results of the previous main process, it achieves rapid convergence in the search for strong lineups in the new version, ensuring that test results can respond promptly to game changes and gradually approach the true optimal solution.

[0134] The method in this embodiment involves the following association strategies during application: 1. Multidimensional Abstraction Modeling System It constructs a lineup data structure that supports dynamic expansion and provides a unified configuration interface, allowing for flexible addition or removal of elements to adjust values; a multi-level evaluation index system that supports linear combinations and weight configurations of strength indicators such as win rate and kill ratio with balance indicators such as diversity and adaptability; and a decoupled battle simulator architecture that separates the game rule engine from the balance testing algorithm, improving cross-game pluggable reusability.

[0135] 2. Improved Genetic Algorithm Search Strategy An alternating front / back row pruning evolutionary mechanism is employed, restricting effective evolutionary points in each round to a single row, thus reducing the search space while promoting the emergence of suboptimal solutions. A similarity-weighted RTS steady-state replacement strategy is introduced, maintaining population diversity through nearest-neighbor competition. This avoids premature convergence to local optima and provides more comprehensive gameplay test results. An additional tournament mechanism generates the next round's challenge set by mixing random samples from the parent generation with the current round's top performers, improving evaluation robustness and suppressing overfitting to fixed opponents.

[0136] 3. Multi-dimensional evaluation pipeline and real-time benchmark strength update mechanism Heuristic hierarchical evaluation uses small-scale tournaments for rapid scoring during the search phase and then upgrades to a more rigorous round-robin evaluation scenario during the candidate convergence phase, balancing computational efficiency and testing accuracy. A rolling update strategy for the benchmark lineup is designed, based on novelty verification and formation coverage requirements, to dynamically refresh the benchmark set from the top lineups in each round, ensuring that the strength of the benchmark lineup is updated in real time.

[0137] This embodiment employs reinforcement learning algorithms such as PPO to construct a policy network, which autonomously learns the optimal lineup configuration (such as general, skill, and equipment combinations) in an end-to-end manner through interaction with the simulation environment. Theoretically, this method has the ability to capture complex lineup design patterns, but in practical applications, it faces challenges such as complex reward function design, poor training stability, and high computational consumption, making stable convergence difficult with limited resources. A game type knowledge base can be constructed at the abstract modeling layer, using pre-trained models to achieve parameter mapping, thereby accelerating the adaptation process for new games. However, this approach relies on a large amount of labeled data for domain adaptation training, resulting in high initial implementation costs, and its effectiveness is significantly affected by the quality of historical data and the similarity to game mechanics.

[0138] The distributed extended combat simulator approach improves testing throughput by deploying distributed computing frameworks like Ray to run massive amounts of combat simulations in parallel. However, at the problem scale addressed in this invention, the pruning strategy and hierarchical evaluation mechanism of genetic algorithms can effectively control computational power consumption; the distributed solution introduces higher deployment and scheduling complexity, resulting in relatively limited cost-effectiveness, and is more suitable as a supplementary optimization direction for future significant increases in problem scale.

[0139] The method of this disclosure has the following beneficial effects: 1. Balancing search efficiency and accuracy: By employing crossover, mutation, and survival of the fittest in genetic algorithms, combined with pre / post ranking domain constraints and illegal solution pruning, and alternating use of round-robin / tournament evaluation during the evolutionary stage, the search converges quickly along the near-optimal path. Compared to insufficient coverage of manual testing and lengthy A / B testing cycles, this method significantly compresses the search space and number of simulations while maintaining evaluation accuracy. In typical scenarios, the computational power requirement can be reduced by about 95%, and the limitations of reinforcement learning in terms of difficulty in reward shaping and unstable training are avoided.

[0140] 2. Techniques for maintaining outcome diversity: This solution introduces a "nearest neighbor competition" mechanism, utilizing RTS steady-state replacement and similarity constraints to effectively suppress over 80% of similar homogeneous solutions and premature convergence. It not only preserves diverse and excellent solutions over the long term but also enhances the exploration coverage of the solution space. Therefore, the solution produces a richer set of candidate solutions, enabling a comprehensive evaluation of game balance under different strategies, thus significantly reducing the risk of imbalance after launch. 3. Dynamically adapt to game iterations: Using the top lineups from previous rounds as a starting point, the system can quickly approach the optimal lineups based on existing knowledge during version updates, increasing the search speed for strong lineups by approximately 90%. Combined with the modular decoupling of the battle simulator, evaluation scenarios, and lineup structure (supporting the addition and removal of elements such as equipment and mounts), it still has computability and low-cost migration capabilities for expansion mechanisms such as 3v3 to 5v5.

[0141] 4. Enhance interpretability: It integrates interactive visualization, allowing filtering by formation, front / back row, heroes, and other criteria, and aggregates and deduplicates similar lineups. It provides a multi-faceted breakdown of formation strength (e.g., a lineup is strong because of its formation strength rather than the strength of its heroes); the results are intuitive and easy for designers to quickly compare and adjust parameters for testing.

[0142] Figure 10 This is a block diagram illustrating a service attribute updating device according to an exemplary embodiment. (Refer to...) Figure 10 The device includes: The lineup attribute acquisition module 1010 is configured to acquire the lineup attribute information set corresponding to each seed lineup in the seed lineup pool of the preset business; the lineup attribute information set includes at least two lineup attribute information sets. The lineup attribute classification module 1020 is configured to perform a lineup attribute information set for each seed lineup, and divide the lineup attribute information set into front-row lineup attribute information and back-row lineup attribute information according to the attribute type corresponding to each lineup attribute information, so as to obtain the front-row lineup attribute information and back-row lineup attribute information corresponding to each seed lineup. The lineup pool update module 1030 is configured to sequentially use the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup as the fixed attribute information of each seed lineup; fix the fixed attribute information of each seed lineup in the seed lineup pool, and iteratively update the non-fixed attribute information of each seed lineup to obtain an updated lineup pool; the non-fixed attribute information of each seed lineup is the attribute information other than the fixed attribute information in the front-row lineup attribute information and the back-row lineup attribute information corresponding to each seed lineup. The target lineup determination module 1040 is configured to determine the target lineup based on the battle results between lineups in the updated lineup pool.

[0143] In one exemplary implementation, the lineup pool update module includes: The attribute classification unit is configured to take the front row lineup attribute information or the back row lineup attribute information corresponding to each seed lineup as the first lineup attribute information of each seed lineup; and to determine the attribute information other than the first lineup attribute information in the front row lineup attribute information and the back row lineup attribute information corresponding to each seed lineup as the second lineup attribute information of each seed lineup. The initial update unit is configured to fix the second lineup attribute information of each seed lineup in the seed lineup pool, and iteratively update the first lineup attribute information of each seed lineup to obtain an initial updated lineup pool; the initial updated lineup pool includes the second lineup attribute information and the first updated lineup attribute information corresponding to each initial updated lineup; the first updated lineup attribute information is the attribute information after the first lineup attribute information is updated. The lineup pool construction unit is configured to construct the updated lineup pool based on the battle results between the initial updated lineup in the initial updated lineup pool and the seed lineup in the seed lineup pool.

[0144] In one exemplary implementation, the lineup pool construction unit includes: The battle result acquisition subunit is configured to acquire the parent-child battle results between each initial updated lineup in the initial updated lineup pool and each seed lineup in the seed lineup pool; The lineup selection subunit is configured to determine a selection pool from the initial updated lineup pool and the seed lineup pool based on the battle results. The iterative subunit is configured to fix the first lineup attribute information of each lineup in the lineup pool, and iteratively update the second lineup attribute information of each lineup to obtain the second updated lineup attribute information. The lineup construction subunit is configured to construct an updated lineup based on the first lineup attribute information and the second updated lineup attribute information corresponding to each filtered lineup; and to construct the updated lineup pool based on the updated lineup; if the filtered lineup is the initial updated lineup, the first lineup attribute information of the filtered lineup is the first updated lineup attribute information; if the filtered lineup is the seed lineup, the first lineup attribute information of the filtered lineup is the first lineup attribute information of the seed lineup.

[0145] In one exemplary implementation, the iterative subunit includes being configured to: fix the second lineup attribute information of each seed lineup in the seed lineup pool; perform a lineup crossover operation based on the first lineup attribute information of each seed lineup to obtain a first updated lineup; perform a lineup mutation operation based on the first lineup attribute information of each seed lineup to obtain a second updated lineup; and construct the initial updated lineup pool based on the first updated lineup and the second updated lineup.

[0146] In one exemplary implementation, the battle result acquisition subunit is further configured to acquire each initial updated lineup in the initial updated lineup pool as a parent lineup; acquire a seed lineup in the seed lineup pool as a child lineup; and determine the battle result of the parent and child lineups based on the win rate, survival rate, and loss ratio of the parent and child lineups in a preset business scenario.

[0147] In one exemplary implementation, the battle result acquisition subunit is further configured to acquire the win rate, survival rate, and kill ratio of the parent lineup and the child lineup in a round-robin scenario to obtain a first battle result; and acquire the win rate, survival rate, and kill ratio of the parent lineup and the child lineup in a tournament scenario to obtain a second battle result; and determine the parent-child lineup battle result based on the first battle result and the second battle result.

[0148] In one exemplary implementation, the lineup selection subunit includes: The first similarity determination subunit is configured to perform the following: determine the character similarity, skill similarity, and formation similarity between the parent lineups based on the lineup attribute information corresponding to the parent lineups in the seed lineup pool. The difference result determination subunit is configured to determine the lineup difference results of the seed lineup pool based on the character similarity, skill similarity, and formation similarity between the parent lineups; The second similarity determination subunit is configured to perform the following: determine the character similarity, skill similarity, and formation similarity between the child lineups based on the lineup attribute information corresponding to the child lineups in the initial updated lineup pool. The lineup difference determination subunit is configured to determine the lineup difference results of the initial updated lineup pool based on the character similarity, skill similarity, and formation similarity between the child lineups; The filtering subunit is configured to determine a filtering lineup from the initial updated lineup pool and the seed lineup pool based on the battle results, the lineup difference results of the seed lineup pool, and the lineup difference results of the initial updated lineup pool.

[0149] In one exemplary implementation, the filtering subunit is configured to perform the following actions: determining a first battle score for the seed lineup pool and a second battle score for the initial updated lineup pool based on the battle results; determining a comprehensive evaluation score for the seed lineup pool based on the first battle score and the lineup difference results of the seed lineup pool; determining a comprehensive evaluation score for the initial updated lineup pool based on the second battle score and the lineup difference results of the initial updated lineup pool; and determining a selection lineup from the initial updated lineup pool and the seed lineup pool based on the comprehensive evaluation scores of the seed lineup pool and the initial updated lineup pool.

[0150] In one exemplary implementation, the lineup difference determination subunit is configured to perform the following operations: determine a first difference result based on the character similarity between the parent lineups; determine a second difference result based on the skill similarity between the parent lineups; determine a third difference result based on the formation similarity between the parent lineups; obtain a first weight corresponding to the character similarity, a second weight corresponding to the skill similarity, and a third weight corresponding to the formation similarity; and determine the lineup difference result of the seed lineup pool based on the product of the first difference result and the first weight, the product of the second difference result and the second weight, and the product of the third difference result and the third weight.

[0151] In one exemplary implementation, the lineup pool construction unit includes: The battle result acquisition subunit is configured to acquire the battle results between each initial updated lineup in the initial updated lineup pool. The candidate lineup construction subunit is configured to perform the task of selecting candidate lineups from the initial updated lineup pool based on the battle results between each initial updated lineup, and constructing a candidate lineup pool. The lineup pool construction subunit is configured to construct the updated lineup pool based on the battle results between the candidate lineup pool and the seed lineups in the seed lineup pool, using a restricted tournament replacement strategy.

[0152] In one exemplary implementation, the target lineup determination module includes: The lineup combination building unit is configured to build at least two lineup combinations based on the updated lineup pool; The battle result acquisition unit is configured to acquire the battle results corresponding to each lineup combination in at least two preset business scenarios. The lineup selection unit is configured to select the target lineup from the updated lineup pool based on the battle results of each lineup combination in at least two preset business scenarios. The device further includes: The lineup attribute sending module is configured to send the lineup attribute information of the target lineup to the terminal so that the terminal can display the lineup attribute information of the target lineup.

[0153] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0154] In one exemplary embodiment, an electronic device is also provided, including a processor; a memory for storing processor-executable instructions; wherein, when the processor is configured to execute the instructions stored in the memory, it implements the business attribute update method provided in any of the above embodiments.

[0155] The electronic device can be a terminal, a server, or a similar computing device. Taking a server as an example... Figure 11 This is a block diagram of a server according to an exemplary embodiment, such as... Figure 11 As shown, the server 1100 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 1110 (CPUs 1110 may include, but are not limited to, microprocessors (MCUs) or programmable logic devices (FPGAs), a memory 1130 for storing data, and one or more storage media 1120 (e.g., one or more mass storage devices) for storing application programs 1123 or data 1122. The memory 1130 and storage media 1120 may be temporary or persistent storage. The program stored in the storage media 1120 may include one or more modules, each module may include a series of instruction operations on the server. Furthermore, the CPU 1110 may be configured to communicate with the storage media 1120 and execute the series of instruction operations stored in the storage media 1120 on the server 1100. Server 1100 may also include one or more power supplies 1160, one or more wired or wireless network interfaces 1150, one or more input / output interfaces 1140, and / or one or more operating systems 1121, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0156] The input / output interface 1140 can be used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of server 1100. In one example, the input / output interface 1140 includes a network interface controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the input / output interface 1140 may be a radio frequency (RF) module used for wireless communication with the Internet.

[0157] Those skilled in the art will understand that Figure 11 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, server 1100 may also include... Figure 11 The more or fewer components shown, or having the same Figure 11 The different configurations shown.

[0158] In one exemplary embodiment, a computer-readable storage medium including instructions is also provided, such as a memory 1130 including instructions, which can be executed by a processor 1110 of server 1100 to perform the above-described method. Optionally, the computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0159] In one exemplary embodiment, a computer program product is also provided, including a computer program that, when executed by a processor, implements the business attribute update method provided in any of the above embodiments.

[0160] Figure 12 This is a block diagram illustrating an electronic device for updating service attributes according to an exemplary embodiment. The electronic device may be a terminal, and its internal structure diagram may be as follows: Figure 12As shown, the electronic device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a business attribute update method. The display screen can be a liquid crystal display (LCD) or an e-ink display. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the device's casing, or an external keyboard, touchpad, or mouse. Those skilled in the art will understand that Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present disclosure and does not constitute a limitation on the electronic device to which the present disclosure is applied. A specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0161] In an exemplary embodiment, an electronic device is also provided, comprising: A processor; a memory for storing processor-executable instructions; wherein the processor is configured to execute the instructions to implement the aforementioned business attribute update method.

[0162] In an exemplary embodiment, a computer-readable storage medium including instructions is also provided, such as a memory including instructions, which can be executed by a processor of an electronic device to complete the aforementioned service attribute update method. Optionally, the computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0163] In an exemplary embodiment, a computer program product is also provided, including a computer program that, when executed by a processor, implements the above-described business attribute update method.

[0164] This disclosure obtains a set of lineup attribute information corresponding to each seed lineup in a seed lineup pool for a preset service; the set of lineup attribute information includes at least two sets of lineup attribute information; for each set of lineup attribute information corresponding to a seed lineup, the set of lineup attribute information is divided into front-row lineup attribute information and back-row lineup attribute information according to the attribute type corresponding to each set of lineup attribute information, to obtain the front-row lineup attribute information and back-row lineup attribute information corresponding to each seed lineup; then, the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup is sequentially used as the fixed attribute information of each seed lineup; the fixed attribute information of each seed lineup in the seed lineup pool is fixed. The system collects and iteratively updates the non-fixed attribute information of each seed lineup to obtain an updated lineup pool. The non-fixed attribute information of each seed lineup includes the attribute information of the front-row lineup and the attribute information of the back-row lineup, excluding the fixed attribute information. By fixing the front-row or back-row attributes in the lineup, the search space is effectively limited to a few positions, significantly reducing the difficulty of searching for high-quality lineups and promoting the generation of diverse solutions within a single row domain. On the one hand, it significantly reduces the search space; on the other hand, searching within a single row domain (front-row lineup attribute information or back-row lineup attribute information) is more likely to produce diverse suboptimal results, ensuring lineup diversity and reducing the risk of premature convergence. Finally, based on the battle results between lineups in the updated lineup pool, the target lineup is determined. This can increase the probability that the selected target lineup is a high-quality lineup and increase the diversity of the selected target lineups.

[0165] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0166] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0167] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A service attribute update method characterized by comprising: The method comprises the following steps: obtaining a set of lineup attribute information corresponding to each seed lineup in a preset seed lineup pool of a service; the set of lineup attribute information comprises at least two pieces of lineup attribute information; for the set of lineup attribute information corresponding to each seed lineup, the set of lineup attribute information is divided into front-row lineup attribute information and back-row lineup attribute information according to the attribute types corresponding to each piece of lineup attribute information, so as to obtain the front-row lineup attribute information and the back-row lineup attribute information corresponding to each seed lineup; the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup is sequentially taken as fixed attribute information of each seed lineup; the fixed attribute information of each seed lineup in the seed lineup pool is fixed, and non-fixed attribute information of each seed lineup is iteratively updated to obtain an updated lineup pool; the non-fixed attribute information of each seed lineup is attribute information other than the fixed attribute information in the front-row lineup attribute information and the back-row lineup attribute information corresponding to each seed lineup; determining a target lineup based on a battle result between lineups in the updated lineup pool.

2. The method of claim 1, wherein, The method of sequentially taking the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup as fixed attribute information of each seed lineup, fixing the fixed attribute information of each seed lineup in the seed lineup pool, and iteratively updating the non-fixed attribute information of each seed lineup to obtain an updated lineup pool comprises the following steps: taking the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup as first lineup attribute information of each seed lineup; and determining attribute information other than the first lineup attribute information in the front-row lineup attribute information and the back-row lineup attribute information corresponding to each seed lineup as second lineup attribute information of each seed lineup; fixing the second lineup attribute information of each seed lineup in the seed lineup pool, and iteratively updating the first lineup attribute information of each seed lineup to obtain an initial updated lineup pool; the initial updated lineup pool comprises the second lineup attribute information and first updated lineup attribute information corresponding to each initial updated lineup; the first updated lineup attribute information is attribute information updated from the first lineup attribute information; constructing the updated lineup pool based on a battle result between an initial updated lineup in the initial updated lineup pool and a seed lineup in the seed lineup pool.

3. The method of claim 2, wherein, The method of constructing the updated lineup pool based on a battle result between an initial updated lineup in the initial updated lineup pool and a seed lineup in the seed lineup pool comprises the following steps: obtaining a parent-child lineup battle result between each initial updated lineup in the initial updated lineup pool and each seed lineup in the seed lineup pool; determining a screening lineup pool from the initial updated lineup pool and the seed lineup pool according to the battle result; fixing first lineup attribute information of each screening lineup in the screening lineup pool, and iteratively updating second lineup attribute information of each screening lineup to obtain second updated lineup attribute information; and Construct an updated lineup based on the first lineup attribute information and the second updated lineup attribute information corresponding to each screening lineup, and construct the updated lineup pool based on the updated lineup.

4. The method of claim 2, wherein, The method further includes the following steps of: fixing the second lineup attribute information of each seed lineup in the seed lineup pool, and iteratively updating the first lineup attribute information of each seed lineup to obtain an initial updated lineup pool, including: fixing the second lineup attribute information of each seed lineup in the seed lineup pool, and performing a lineup crossover operation based on the first lineup attribute information of each seed lineup to obtain a first updated lineup; performing a lineup mutation operation based on the first lineup attribute information of each seed lineup to obtain a second updated lineup; 5. The method of claim 3, wherein, constructing the initial updated lineup pool based on the first updated lineup and the second updated lineup. The method further includes the following steps of: obtaining the parent-child lineup battle result between each initial updated lineup in the initial updated lineup pool and each seed lineup in the seed lineup pool, including:

6. The method of claim 5, wherein, obtaining each initial updated lineup in the initial updated lineup pool as a parent lineup, and obtaining a seed lineup in the seed lineup pool as a child lineup; determining the parent-child lineup battle result according to the winning rate, survival rate and battle damage ratio of the parent lineup and the child lineup in a preset business scenario. The method further includes the following steps of: obtaining the winning rate, survival rate and battle damage ratio of the parent lineup and the child lineup in a round robin scenario to obtain a first battle result; 7. The method according to any one of claims 3-6, characterized in that, obtaining the winning rate, survival rate and battle damage ratio of the parent lineup and the child lineup in a tournament scenario to obtain a second battle result; determining the parent-child lineup battle result according to the first battle result and the second battle result. The method further includes the following steps of: determining the role similarity, skill similarity and formation similarity between the parent lineups according to the lineup attribute information corresponding to the parent lineups in the seed lineup pool; determining the lineup difference result of the seed lineup pool according to the role similarity, skill similarity and formation similarity between the parent lineups; determining the role similarity, skill similarity and formation similarity between the child lineups according to the lineup attribute information corresponding to the child lineups in the initial updated lineup pool; 8. The method of claim 7, wherein, determining the lineup difference result of the initial updated lineup pool according to the role similarity, skill similarity and formation similarity between the child lineups; determining the screening lineups from the initial updated lineup pool and the seed lineup pool based on the battle result, the lineup difference result of the seed lineup pool and the lineup difference result of the initial updated lineup pool. The method further includes the following steps of: determining the role similarity, skill similarity and formation similarity between the parent lineups according to the lineup attribute information corresponding to the parent lineups in the seed lineup pool; determining the lineup difference result of the seed lineup pool according to the role similarity, skill similarity and formation similarity between the parent lineups; determining the role similarity, skill similarity and formation similarity between the child lineups according to the lineup attribute information corresponding to the child lineups in the initial updated lineup pool; determining the lineup difference result of the initial updated lineup pool according to the role similarity, skill similarity and formation similarity between the child lineups; determining the screening lineups from the initial updated lineup pool and the seed lineup pool based on the battle result, the lineup difference result of the seed lineup pool and the lineup difference result of the initial updated lineup pool. determine a first battle score of the seed lineup pool and a second battle score of the initial updated lineup pool based on the battle results; determine a comprehensive evaluation score of the seed lineup pool based on the first battle score and a lineup difference result of the seed lineup pool; determine a comprehensive evaluation score of the initial updated lineup pool based on the second battle score and a lineup difference result of the initial updated lineup pool; determine a screening lineup from the initial updated lineup pool and the seed lineup pool according to the comprehensive evaluation score of the seed lineup pool and the comprehensive evaluation score of the initial updated lineup pool.

9. The method of claim 7, wherein, The determining of the lineup difference result of the seed lineup pool according to the role similarity, the skill similarity and the formation similarity between the parent lineups comprises: determining a first difference result according to the role similarity between the parent lineups; determining a second difference result according to the skill similarity between the parent lineups; determining a third difference result according to the formation similarity between the parent lineups; obtaining a first weight corresponding to the role similarity, a second weight corresponding to the skill similarity and a third weight corresponding to the formation similarity; determining the lineup difference result of the seed lineup pool according to the product of the first difference result and the first weight, the product of the second difference result and the second weight and the product of the third difference result and the third weight.

10. The method of claim 2, wherein, The constructing of the updated lineup pool based on the battle results between the initial updated lineups in the initial updated lineup pool and the seed lineups in the seed lineup pool comprises: obtaining the battle results between each initial updated lineup in the initial updated lineup pool; screening out candidate lineups from the initial updated lineup pool according to the battle results between each initial updated lineup to construct a candidate lineup pool; constructing the updated lineup pool by using a restricted tournament replacement strategy according to the battle results between the candidate lineup pool and the seed lineups in the seed lineup pool.

11. The method of claim 1, wherein, The determining of the target lineup based on the battle results between the lineups in the updated lineup pool comprises: constructing at least two lineup combinations according to the updated lineup pool; obtaining respective battle results of each lineup combination in at least two preset business scenarios; screening out the target lineup from the updated lineup pool according to the respective battle results of each lineup combination in at least two preset business scenarios. The method further comprises: sending lineup attribute information of the target lineup to a terminal to enable the terminal to display the lineup attribute information of the target lineup.

12. A service attribute update apparatus characterized by comprising: comprises: a lineup attribute obtaining module configured to obtain a set of lineup attribute information corresponding to each seed lineup in a seed lineup pool of a preset business; the set of lineup attribute information comprises at least two pieces of lineup attribute information; a lineup attribute classifying module configured to divide, for each seed lineup, the set of lineup attribute information corresponding to the seed lineup into front-row lineup attribute information and back-row lineup attribute information according to an attribute type corresponding to each piece of lineup attribute information, to obtain the front-row lineup attribute information and the back-row lineup attribute information corresponding to each seed lineup; The lineup pool updating module is configured to sequentially take the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup as fixed attribute information of each seed lineup; fix the fixed attribute information of each seed lineup in the seed lineup pool, and iteratively update the non-fixed attribute information of each seed lineup to obtain an updated lineup pool; the non-fixed attribute information of each seed lineup is attribute information other than the fixed attribute information in the front-row lineup attribute information or the back-row lineup attribute information corresponding to each seed lineup; The target lineup determination module is configured to determine a target lineup based on a battle result between lineups in the updated lineup pool.

13. An electronic device, comprising: Comprise: A processor; A memory for storing instructions executable by the processor; Wherein the processor is configured to execute the instructions to implement the business attribute updating method of any one of claims 1-11.

14. A computer-readable storage medium, characterized in that, When the instructions in the computer readable storage medium are executed by the processor of the electronic device, the electronic device can execute the business attribute updating method of any one of claims 1-11.

15. A computer program product comprising computer instructions, characterized in that, The computer instructions are executed by the processor to implement the business attribute updating method of any one of claims 1-11.

Images