Data processing method and device, equipment, storage medium and product

Abstract

The invention provides a data processing method and device, equipment, a storage medium and a product, and the data processing method comprises the steps: obtaining to-be-evaluated content and a plurality of target digital twin bodies, and enabling the target digital twin bodies to be pre-constructed based on the user historical data of a target real user; according to the to-be-assessed content and the digital twinborn information of each target digital twinborn, generating assessment information of each target digital twinborn for assessing the to-be-assessed content; and obtaining an assessment result for the to-be-assessed content according to each piece of assessment information. The method can be widely applied to the fields of digital culture product manufacturing software, digital culture creative software and the like in a new generation of information technology.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a data processing method, apparatus, device, storage medium, and product. Background Technology

[0002] In the fields of digital cultural product creation software and digital cultural creative software, it is possible to develop and operate related business software such as game software. This development or operation involves things like game launches and updates.

[0003] Generally, it's necessary to obtain player feedback on game updates to evaluate user experience. This can usually be done through methods like setting up test servers or conducting surveys. However, because different players have different personalities, not all players will participate in test servers or fill out surveys. For example, extroverted players are more likely to fill out surveys, while introverted players are less likely to. This results in biased feedback, making it difficult to obtain accurate and comprehensive player feedback on game updates. Summary of the Invention

[0004] In view of this, embodiments of this application provide a data processing method. This application also relates to a digital twin construction method, a data processing apparatus, a computing device, a computer-readable storage medium, and a computer program product, to solve the aforementioned problems existing in the prior art.

[0005] According to a first aspect of the embodiments of this application, a data processing method is provided, including: Acquire the content to be evaluated and multiple target digital twins, wherein the target digital twins are pre-built based on the user history data of the target real users; Based on the content to be evaluated and the digital twin information of each target digital twin, evaluation information is generated for each target digital twin to evaluate the content to be evaluated. The evaluation results for the content to be evaluated are obtained based on the evaluation information.

[0006] According to a second aspect of the embodiments of this application, a method for constructing a digital twin is provided, comprising: Obtain user history data from multiple real users; Based on multiple preset data dimensions and the user history data of each real user, generate digital twin information corresponding to each real user; Based on the digital twin information corresponding to each real user, obtain the digital twin corresponding to each real user.

[0007] According to a third aspect of the embodiments of this application, a data processing apparatus is provided, comprising: The acquisition module is configured to acquire the content to be evaluated and multiple target digital twins, wherein the target digital twins are pre-built based on the user history data of the target real users; The generation module is configured to generate evaluation information for each target digital twin to evaluate the content to be evaluated, based on the content to be evaluated and the digital twin information of each target digital twin. The acquisition module is configured to obtain the evaluation results for the content to be evaluated based on the evaluation information.

[0008] According to a fourth aspect of the embodiments of this application, a computing device is provided, comprising: Memory and processor; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions, which, when executed by the processor, implement the steps of the above-described data processing method.

[0009] According to a fifth aspect of the embodiments of this application, a computer-readable storage medium is provided that stores a computer program / instructions, which, when executed by a processor, implement the steps of the above-described data processing method.

[0010] According to a sixth aspect of the embodiments of this application, a computer program product is provided, including a computer program / instructions that, when executed by a processor, implement the steps of the above-described data processing method.

[0011] The data processing method provided in this application can obtain the content to be evaluated and multiple target digital twins, wherein the target digital twins are pre-constructed based on the user history data of the target real users; according to the content to be evaluated and the digital twin information of each target digital twin, evaluation information is generated for each target digital twin to evaluate the content to be evaluated; and evaluation results are obtained for the content to be evaluated based on the evaluation information.

[0012] One embodiment of this application constructs a digital twin based on the user history data of real users and performs an evaluation on the content to be evaluated. Since the digital twin is constructed based on the user history data of real users, it is a mapping of real users. Therefore, by using the digital twin to evaluate the content to be evaluated, the situation where real users do not participate in the evaluation of the content to be evaluated can be avoided, and accurate and comprehensive evaluation results can be obtained. Attached Figure Description

[0013] Figure 1This is a flowchart illustrating a digital twin construction method provided in one embodiment of this specification; Figure 2 This is a flowchart illustrating a data processing method provided in one embodiment of this specification; Figure 3 This is a schematic diagram of a starting target digital twin and an associated target digital twin provided in one embodiment of this specification; Figure 4 This is a schematic diagram of the structure of a data processing apparatus provided in one embodiment of this specification; Figure 5 This is a structural block diagram of a computing device provided in one embodiment of this specification. Detailed Implementation

[0014] Many specific details are set forth in the following description to provide a full understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application; therefore, this application is not limited to the specific embodiments disclosed below.

[0015] The terminology used in one or more embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the scope of one or more embodiments of this application. The singular forms “a,” “the,” and “the” used in one or more embodiments of this application and in the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” used in one or more embodiments of this application refers to and includes any or all possible combinations of one or more associated listed items.

[0016] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this application, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this application, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0017] 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 analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0018] First, the terms and concepts involved in one or more embodiments of this application will be explained.

[0019] Lifetime Value (LTV): Also known as customer lifetime value or user lifetime value. LTV can be understood as the total expected net revenue a user brings to a business or organization over their entire lifespan. LTV is one of the core metrics for quantifying user value.

[0020] As described in the background section, the evaluation of game updates in the gaming industry is crucial. However, current evaluations are often delayed and irreversible. Specifically, user feedback on game updates, as well as data changes and public opinion regarding the updates, can only be obtained after the update is officially released or has been experienced by players for a period of time. Furthermore, setting up test servers or conducting surveys cannot cover all players; typically, only a select group or a small number of core players participate in test servers or surveys, leading to inaccurate and incomplete evaluations of game updates.

[0021] Based on this, this specification provides a data processing method that can acquire the content to be evaluated and multiple target digital twins, wherein the target digital twins are pre-constructed based on the user history data of the target real users; according to the content to be evaluated and the digital twin information of each target digital twin, evaluation information is generated for each target digital twin to evaluate the content to be evaluated; and evaluation results for the content to be evaluated are obtained based on the evaluation information.

[0022] By constructing digital twins based on real users' historical data and then evaluating the content to be evaluated, the digital twins are a mapping of real users. Therefore, by using digital twins to evaluate the content to be evaluated, the situation where real users do not participate in the evaluation of the content to be evaluated can be avoided, and accurate and comprehensive evaluation results can be obtained.

[0023] It should be noted that a digital twin can be understood as a high-fidelity, real-time synchronized virtual model of an entity in the physical world, such as equipment, systems, factories, human bodies, cities, etc., built in digital virtual space. This virtual model can reflect the entity's state, behavior, and performance, and support prediction, optimization, and decision-making. In one or more embodiments of this specification, the digital twin can be a digital human twin. A digital human twin can be understood as a digital model or copy of a real human individual constructed in digital virtual space using digital twin technology. This model can reflect the individual's state, behavior, physiological indicators, psychological characteristics, and even social relationships in the physical world in real time, and can be dynamically updated, simulated, predicted, and interactively fed back through data-driven processes.

[0024] In one or more embodiments of this specification, the digital twin is pre-built based on the user history data of real users. Here, "real users" can be understood as real individuals in the physical world. Real users and their historical data differ in different scenarios. In other words, the data processing methods and digital twin construction methods provided in this specification are applicable to different business scenarios and can be used for simulation evaluation of content to be evaluated in various business scenarios.

[0025] For example, in a game scenario, a real user can be a real game user, and the content to be evaluated can be game update content. Game update content includes, but is not limited to, adding game equipment, virtual characters, game maps, gameplay, paid gift packs, or system rule changes, etc. A digital twin can be built based on the historical game data of real game players. Historical game data includes, but is not limited to, game interaction logs, battle records, consumption orders, social relationships, and character attributes, etc.

[0026] For example, in an e-commerce scenario, real users can be real e-commerce users, and the content to be evaluated can be e-commerce content launched online. New e-commerce content includes, but is not limited to, new products, new shopping activity strategies, and adjustments to the e-commerce page format. The digital twin can be built based on the historical data of real e-commerce users, which includes, but is not limited to, browsing history, purchase history, add-to-cart history, product reviews, and follow information.

[0027] Of course, it can also be applied to social media platforms, such as evaluating new feature launches and adjustments to community rules, as well as digital content platforms, such as adjusting short video page display strategies and changing membership benefits. In short, the content to be evaluated, digital twins, real users, and historical user data of real users in this specification can all be determined based on actual scenarios and actual needs, and this specification does not impose specific limitations.

[0028] In one or more embodiments of this specification, a digital twin may be constructed before implementing the data processing method provided in this specification, and the content to be evaluated may be evaluated based on the constructed digital twin.

[0029] like Figure 1 As shown, Figure 1 The following is a flowchart illustrating a digital twin construction method provided in one embodiment of this specification, specifically including the following steps: Step 102: Obtain user history data from multiple real users.

[0030] It should be noted that the execution entity for the digital twin construction method provided in this specification can be any computing device with computing capabilities, such as a server or terminal, etc., and this specification does not impose specific restrictions. Furthermore, the execution entity for the data processing method provided in this specification can be the same as or different from the execution entity for the digital twin construction method, and this specification does not impose specific restrictions.

[0031] As mentioned above, real users and their historical data can be determined based on actual scenarios and needs. For ease of description, in one or more embodiments of this specification, a game scenario is used as an example to illustrate the technical solution. Here, the content to be evaluated is game update content, real users are real game players, and the digital twin is a digital twin of the game player constructed based on the historical game data of the real game player. However, it should be understood that this example is for illustrative purposes only and should not be considered a limitation on the scope of protection of this invention.

[0032] In practical applications, historical game data from real players can be obtained from game log databases. Specifically, if the game update being evaluated is an update to the current game content, or if the game is already live, the historical game data of real players in that current game can be directly retrieved. Alternatively, the historical game data of real players in related games can also be retrieved. If the game update being evaluated is a newly released game, related games can be identified, and the historical game data of real players in those related games can be obtained. The target game can be understood as the game corresponding to the update being evaluated, and related games can be understood as already released games similar to the target game.

[0033] It should be noted that the number of real game players is multiple. Historical game data can include interaction data, behavioral data, transaction data, and attribute data, etc. Interaction data includes, but is not limited to, chat logs, social relationships, etc. Behavioral data includes, but is not limited to, game duration, game record, churn threshold, etc. Transaction data includes, but is not limited to, historical game orders, LTV, spending preferences, etc. Attribute data includes, but is not limited to, the player's friend list information, guild information, game title, and inventory contents, etc.

[0034] In the process of building digital twins of game players, obtaining historical game data from real game players can provide a data foundation for subsequent steps.

[0035] Step 104: Generate digital twin information for each real user based on multiple preset data dimensions and the user history data of each real user.

[0036] In one or more embodiments of this specification, for each real user, a digital twin can be generated based on multiple preset data dimensions and the user's historical data. In other words, the same digital twin construction operation needs to be performed for each real user. Therefore, for ease of description, any real user is referred to as the real user to be processed, that is, any real game player can be referred to as the real game player to be processed, and then a digital twin of the game player corresponding to the real game player to be processed is constructed.

[0037] In practical applications, information corresponding to multiple preset data dimensions can be extracted from historical game data to obtain digital twin information.

[0038] In one or more embodiments of this specification, the preset multiple data dimensions may include interaction data dimensions, behavioral data dimensions, transaction data dimensions, and attribute data dimensions. Specifically, the interaction data dimension can characterize the social expression and communication style of real users, the behavioral data dimension can characterize the operating habits and psychological resilience of real users, the transaction data dimension can characterize the transaction consumption of real users, and the attribute data dimension can characterize the identity recognition and community affiliation of real users.

[0039] In one or more embodiments of this specification, digital twin information corresponding to each real user is generated based on multiple preset data dimensions and the user history data of each real user, including: The first piece of information is determined based on user history data from the interaction data dimension, wherein the first piece of information represents the emotional information and / or social information of real users; or, The second information is determined based on historical user data from the behavioral data dimension, wherein the second information represents the activity information and / or psychological information of real users; or, The third piece of information is determined based on historical user data from the transaction data dimension, wherein the third piece of information represents the transaction information of a real user; or, The fourth information is determined based on the user's historical data in the attribute information dimension, wherein the fourth information represents the identity information of the real user; Digital twin information is obtained based on at least one of the first information, the second information, the third information, and the fourth information.

[0040] Since different data dimensions are used to extract different categories of player characteristics from historical game data, the information identified by each data dimension, representing different psychological or behavioral aspects of game players such as social interaction, behavioral habits, consumption motivations, and identity affiliation, can reflect the differentiated characteristics of real users in the corresponding data dimensions. As mentioned earlier, the interaction data dimension can be used to reflect the emotional information of real game players, such as emotional expression tendencies, and social information, such as social communication styles. The behavioral data dimension can be used to reflect the activity information of real game players, such as activity patterns, and psychological information, such as frustration tolerance. The transaction data dimension can be used to reflect the transaction information of real game players, such as payment types and transaction motivation information, such as willingness to pay and value-driven types. The attribute data dimension can be used to reflect the identity information of real game players, such as the achievement pursuits and community role positioning of the virtual game character corresponding to the game player.

[0041] Information is extracted from historical data from different data dimensions to form digital twin information used to generate digital twins of game players that are real users, thereby improving the accuracy of the generated digital twins.

[0042] In practical applications, interaction data within user history data can include chat text semantics, speaking frequency, sentiment, and the tightness of social relationship chains. Behavioral data within user history data can include active game periods, single session duration, win rate and rank matching relationship, and churn thresholds determined based on offline behavior after consecutive losses. Transaction data within user history data can include historical order amounts, payment frequency, consumer product preferences (e.g., stat-enhancing or cosmetic items), and Lifetime Value (LTV). Attribute data within user history data can include the contents of a player's inventory, game achievements and titles, rank in a guild or clan, friend list size, and user-submitted extended attributes such as birthday, zodiac sign, and other cultural tags.

[0043] It should be noted that the attribute information represents the attribute information or identity information of the game account corresponding to the real game player, rather than the real attribute information or real identity information of the real game player.

[0044] In one or more embodiments of this specification, multiple preset data dimensions and historical game data of the real game player to be processed can be input into a pre-trained model. This allows the pre-trained model to extract information corresponding to each data dimension from the historical game data of the real game player to be processed, specifically structured text information, to obtain first, second, third, and fourth information corresponding to the real game player to be processed. Then, based on at least one of the first, second, third, and fourth information corresponding to the real game player to be processed, digital twin information corresponding to the real user to be processed can be obtained. This allows for the construction of a digital twin of the real user to be processed based on the digital twin information corresponding to the real user to be processed.

[0045] It should be noted that a pre-trained model can be understood as a machine learning model that has been trained in advance. This pre-trained machine learning model can be a pre-trained language model (such as a large language model), or a pre-trained multimodal model, etc. In the case of a pre-trained language model, the generated digital twin information is textual information.

[0046] Furthermore, in one or more embodiments of this specification, the category label of the real user to be processed can be determined based on the extracted information. Specifically, the first category label of the real game player to be processed can be determined based on the first information corresponding to the real game player to be processed. The first category label represents the classification of the first information (social information, emotional information) of the real game player, and the first category label can be preset. For example, some game players frequently use words such as angry and swear in their chat messages, while some game players frequently use words such as stay calm and cheer up. In this case, the first category label can include categories such as irritable type and calm type for emotional information classification. Similarly, some game players have few or no social chat messages, while some game players have a lot of chat messages. In this case, the first category label can include categories such as talkative type and silent type for social information classification.

[0047] Correspondingly, a second category label can be determined based on the second information corresponding to the real gamers to be processed. This second category label represents the classification of the real gamers' second information (activity information, psychological information), and can be preset. For example, some gamers frequently play games late at night, while others frequently play during the day. The second category label could include classifications of activity information such as activity type within a specified time period or activity type within a specified group. Similarly, some gamers log off immediately after consecutive losses, while others continue playing even after consecutive losses. In this case, the second category label could include classifications of psychological information such as strong psychological resilience or weak psychological resilience.

[0048] Correspondingly, a third category label can be determined for the real game players to be processed based on their third information. This third category label represents the classification of the real game players' third information (transaction information), and it can be pre-set. For example, some game players spend a lot, while others spend little or nothing. The third category label could include categories like "paying players" and "non-paying players" based on transaction information. Furthermore, some game players' transaction types or spending patterns primarily focus on game equipment and combat power, while others' focus is on purchasing in-game cosmetic items and costumes. Therefore, the third category label could also include categories like "winner mentality" and "performance mentality," etc.

[0049] Correspondingly, a fourth category label can be determined based on the fourth information corresponding to the real game player to be processed. This fourth category label represents the classification of the real game player's fourth information (identity information), and can be pre-set. For example, some game players have rare items or many game items or characters in their game accounts, while others only have a few or default game items or characters. Therefore, the fourth category label could include classifications of the player's collected titles, such as collection type or default type. Similarly, some game players hold positions in a guild or organization within the game, while others are simply game visitors without a guild or organization. In this case, the fourth category label could include classifications of the player's guild or organization information, such as guild or organization position categories.

[0050] It should be noted that the correspondence between the aforementioned category labels and information (first, second, third, and fourth information) can be pre-defined; that is, the category labels corresponding to which information can be pre-set. Alternatively, the category labels corresponding to the information in each data dimension can be generated based on the large language model. Or, after pre-setting the correspondence, the correspondence and the information in each data dimension can be input into the large language model, allowing the model to determine the category labels corresponding to the information in each data dimension based on the correspondence and the information itself.

[0051] For example, the first piece of information in the player's historical game data includes: "I'm angry, are you stupid? You can't even beat this." Therefore, the first category label can be determined as the "Angry" type.

[0052] Therefore, in one or more embodiments of this specification, digital twin information corresponding to the real game player to be processed can be obtained based on the information of each data dimension and the category labels corresponding to each data dimension. In practical applications, the digital twin information can be textual descriptions of the information of each data dimension and the category labels. Specifically, a prompt template can be pre-built, which may include four fixed slots, each corresponding one-to-one with the information of the four data dimensions. Then, digital twin information can be generated based on the prompt template, the information of each dimension, and the category labels corresponding to the information of each data dimension.

[0053] In practical applications, digital twin information can be organized into structured data objects or natural language descriptive text based on prompt templates, so as to drive the game player's digital twin to make a simulation response consistent with the digital twin information for the content to be evaluated.

[0054] For example, the prompt template could be: You are a person with [interaction data dimension] + [behavioral data dimension] + [transaction data dimension] + [attribute information dimension]. The generated digital twin information could be: You are easily angered and habitually bossy. In interactions, you frequently use sarcastic language (e.g., "Is that all?", "Can't you even play?") because you believe most people are not worthy of playing with you. You are a seasoned, highly invested, leader-type player. You are the guild leader in the game and ranked in the top 10 across the server. You are an extreme power-hungry player. You not only have money, but you also demand that every penny you spend must grant you dominance. When evaluating new content, you only look at whether it increases your combat stats. If it's a purely cosmetic update, you will mock the developers for "not doing their job properly." You have extremely low mental resilience. For example, if you are shown to have lost three games in a row, or if your powerful artifact is weakened, you must immediately explode and verbally abuse the developers in the public chat, threatening to quit the game with your entire guild.

[0055] Step 106: Obtain the digital twin corresponding to each real user based on the digital twin information of each real user.

[0056] It should be noted that, as mentioned earlier, the digital twin information is stored as the core configuration parameters of its corresponding digital twin, which is used to inject context when calling pre-trained models such as large language models.

[0057] In practical applications, the digital twin of the real game player to be processed can be obtained by associating the digital twin information of the real game player to be processed with the digital twin identifier of the corresponding digital twin of the real game player. In other words, the digital twin information of the digital twin identifier represents the digital twin.

[0058] In one or more embodiments of this specification, a digital twin library can be constructed, that is, the digital twin information corresponding to each real game player can be stored in the digital twin library. In practical applications, an interactive interface can be set up, through which digital twins in the digital twin library can be called, or the content to be evaluated can be sent to the computing device where the digital twin library is located for evaluation.

[0059] By constructing digital twins in the above manner, multiple digital twins can be built using real users' historical data. This allows for the evaluation of the content to be evaluated based on the digital twins, obtaining evaluation information of the content to be evaluated from the digital twins that reflect the real users. Furthermore, it can cover real users with different personalities and traits, improving the accuracy and comprehensiveness of the obtained evaluation information.

[0060] Furthermore, to improve retrieval efficiency and simulation representativeness, feature vectorization, clustering, and continuous learning mechanisms can be introduced.

[0061] In one or more embodiments of this specification, when constructing a digital twin library, feature vectors corresponding to the information of each digital twin can be generated by a large language model, and then the feature vectors, digital twin information, and digital twin identifiers of each digital twin are stored in the digital twin library.

[0062] By storing the feature vectors of digital twins, feature comparison can be performed quickly, allowing for the selection of qualified digital twins for evaluation, thus improving the efficiency and accuracy of the selected or retrieved digital twins. For example, when evaluating new game mechanics, a group of digital twins with strong learning intentions and high adversity quotients can be quickly retrieved, thereby improving the relevance of the selected digital twins to the content being evaluated and enhancing the accuracy of the evaluation results.

[0063] In one or more embodiments of this specification, a pre-defined clustering algorithm can be used to cluster the digital twins based on their information or feature vectors to obtain digital twin groups. Each digital twin group can correspond to a game player prototype, and the pre-defined clustering algorithm can be the K-Means clustering algorithm. Alternatively, digital twins can be grouped based on a pre-defined prototype correspondence, where the pre-defined prototype correspondence can be understood as the correspondence between digital twin information and category labels and game player prototypes.

[0064] It should be noted that the game player prototype represents real gamers with similar personalities and behavioral patterns. Game player prototypes can be pre-set based on different digital twin information and category tags. This manual does not specifically limit what kind of digital twin information and category tags correspond to what kind of game player prototype; it can be set based on actual scenarios and needs. For example, game player prototypes can include: a first game player prototype (casual free-to-play player), characterized by low spending, high social dependence, insensitivity to numerical balance, and activity on weekends; a second game player prototype (irritable heavy spender), characterized by high LTV, pursuit of dominance, low emotional stability, and susceptibility to churn due to dissatisfaction with matchmaking mechanisms; and a third game player prototype (story-driven player), characterized by a preference for single-player games, preference for cosmetic purchases, high collection and achievement motivation, and an artistic writing style.

[0065] By setting up different game player prototypes, when evaluating content based on digital twins, digital twins can be selectively extracted based on the game player prototypes, without using all digital twins. This saves computing resources while ensuring the diversity of the selected digital twins' personalities or information.

[0066] In one or more embodiments of this specification, external knowledge can be acquired at specified intervals. This external knowledge can be understood as information related to the target game corresponding to the content to be evaluated. External knowledge includes, but is not limited to, game information such as new game releases, popular gameplay trends, community opinions, popular comments, internet slang and expression paradigms, etc. Then, based on this external knowledge, the information of each digital twin is dynamically updated. Specifically, for each digital twin, its digital twin information and external knowledge can be input into a large language model. The large language model can adjust the digital twin information based on the external knowledge to obtain adjusted digital twin information, thereby updating relevant information within the digital twin.

[0067] By updating the information of the digital twin based on external knowledge, it is possible to ensure that the digital twin always reflects the current cognitive state and cultural context of the game player, avoid the situation where the simulation evaluation results deviate from reality due to the solidification of the digital twin model, and improve the accuracy of the simulation evaluation.

[0068] In practical applications, after constructing multiple digital twins or digital twin databases, an evaluation of the content to be evaluated can be performed based on the constructed digital twins or digital twin databases.

[0069] This application provides a data processing method, and also relates to a data processing apparatus, a computing device, a computer-readable storage medium, and a computer program product, which will be described in detail in the following embodiments.

[0070] like Figure 2 As shown, Figure 2 The flowchart illustrating a data processing method provided in one embodiment of this specification specifically includes the following steps: Step 202: Obtain the content to be evaluated and multiple target digital twins, wherein the target digital twins are pre-built based on the user history data of the target real users.

[0071] It should be noted that the content to be evaluated, the digital twin, the real users, and the users' historical data are consistent with those mentioned above, and will not be repeated here.

[0072] In one or more embodiments of this specification, a target digital twin can be understood as a digital twin selected from a plurality of pre-constructed digital twins. Specifically, obtaining a plurality of target digital twins includes: Multiple target digital twins are selected from each digital twin in the digital twin library according to a preset sampling strategy. The digital twins in the digital twin library are constructed based on the user history data of real users.

[0073] The preset sampling strategy may include sampling according to the category label of each digital twin. The preset sampling strategy may be random sampling, and this specification does not impose specific restrictions.

[0074] It should be noted that, as mentioned earlier, the digital twin library stores digital twin information, category labels, feature vectors, and game player prototypes. Therefore, when selecting a target digital twin, one can choose the target digital twin from the various digital twins in the digital twin library based on one or more of the digital twin information, category labels, feature vectors, and game player prototypes.

[0075] Specifically, the feature vector of the content to be evaluated can be obtained, and then its similarity can be compared with that of the feature vector of the digital twin. The digital twin with the highest similarity can be selected. Alternatively, a target category label, target game player prototype, or target digital twin information matching the content to be evaluated can be determined, and the digital twin corresponding to the target category label, target game player prototype, or target digital twin information can be used as the target digital twin. Furthermore, a category label ratio or game player prototype ratio can be set, and then the digital twins corresponding to each category label or game player prototype can be extracted according to the ratio as the target digital twin. This specification does not impose specific restrictions on the method of obtaining the target digital twin.

[0076] Step 204: Based on the content to be evaluated and the digital twin information of each target digital twin, generate evaluation information for each target digital twin to evaluate the content to be evaluated.

[0077] In one or more embodiments of this specification, evaluation information is generated based on the content to be evaluated and the digital twin information of each target digital twin, including: The content to be evaluated and the digital twin information of the target digital twin are input into a pre-trained language model, so that the pre-trained language model simulates the evaluation process of the target digital twin on the content to be evaluated based on the digital twin information of the target digital twin, and obtains the evaluation information corresponding to the target digital twin, wherein the target digital twin is any one of the target digital twins.

[0078] In practical applications, the digital twin prototype of the target digital twin and the communication between the target and the evaluation content can be used as input information into a pre-trained language model, such as a large language model. The digital twin information can reflect the social, behavioral, transactional, and identity characteristics of real users. The evaluation content can serve as the simulation evaluation trigger information. Based on its language understanding and generation capabilities, the pre-trained language model can simulate the understanding and decision-making process of the target digital twin in a real-world scenario and output the response content to obtain the evaluation information corresponding to the target digital twin.

[0079] It should be noted that the evaluation information can specifically be evaluation text, which includes, but is not limited to, sentiment text regarding the evaluated content, behavioral intention text regarding the evaluated content, reason text for taking that behavioral intention, and social expression text regarding the evaluated content. In other words, the response content output by the pre-trained language model is the response text, which includes, but is not limited to, sentiment text, behavioral intention text, reason text, and social expression text. For example, assuming the evaluated content is new game equipment, the sentiment text could be: "Looking forward to the release of this new game equipment," the behavioral intention text could be: "I will buy this new game equipment," the reason text for taking that behavioral intention could be: "I think this new game equipment looks good and isn't expensive," and the social expression text regarding the evaluated content could be: "The new game equipment is really good; I'll buy it as soon as it's released."

[0080] Furthermore, to improve the accuracy of the evaluation information generated by the pre-trained language model, the content to be evaluated can be processed first. Specifically, the content to be evaluated and the digital twin information of the target digital twin to be processed are input into the pre-trained language model, including: The content to be evaluated is analyzed according to the preset analysis dimensions, and the corresponding evaluation dimension information of the content to be evaluated is generated. The dimensional information to be evaluated and the digital twin information of the target digital twin to be processed are input into the pre-trained language model.

[0081] It should be noted that the preset analysis dimensions include, but are not limited to, cost analysis, benefit analysis, and environmental analysis. Among them, the cost analysis dimension represents the resource cost that players need to pay, which can be time, money, etc. For example, players need to pay a monetary cost of 328 yuan and a time cost of 2 hours per day. The benefit analysis dimension represents the benefits that players can obtain, such as numerical benefits like a 10% increase in attack power, cosmetic benefits like limited-edition virtual game character skins, and social benefits like server-wide announcements. The environmental analysis dimension represents changes in game rules, such as changes in the balance of combat, such as weakening type A virtual game characters, and changes in game mechanics, such as mandatory team play.

[0082] In practical applications, the content to be evaluated can be game design documents, draft update announcements, numerical configuration tables, etc., corresponding to game update content. Therefore, the content to be evaluated can be analyzed based on cost analysis, benefit analysis, and environmental analysis dimensions to obtain the evaluation dimension information corresponding to the preset analysis dimensions. In other words, information corresponding to the preset analysis dimensions can be extracted from the content to be evaluated as the evaluation dimension information.

[0083] For example, the content to be evaluated is: a new limited-edition collaboration skin. The evaluation dimensions can be: a rare skin with only cosmetic effects, priced at 500 yuan, available for a limited time.

[0084] In one or more embodiments of this specification, the generation of evaluation information corresponding to each target game player digital twin can be determined, thus obtaining the evaluation information corresponding to each target game player digital twin.

[0085] Furthermore, since there are relationships between the digital twins of each target game player, or in other words, the evaluation information corresponding to one game player's digital twin will affect the evaluation information of the associated target game player's digital twin, the evaluation results depend not only on the independent simulation evaluation of the digital twins, but also on the correlation simulation evaluation between the digital twin groups. Therefore, in one or more embodiments of this specification, the evaluation information corresponding to each target game player's digital twin can also be obtained based on the relationships between the digital twins of each target game player.

[0086] Specifically, based on the content to be evaluated and the digital twin information of each target digital twin, evaluation information is generated for each target digital twin to evaluate the content to be evaluated, including: Obtain the relationships between the target digital twins; Based on the aforementioned relationships, the content to be evaluated, and the digital twin information of each target digital twin, a social communication simulation is performed on the content to be evaluated to obtain the evaluation information of each target digital twin on the content to be evaluated.

[0087] It should be noted that the relationships can be determined by a pre-trained language model based on the digital twin information of each target game player's digital twin. Specifically, the relationships can be determined by the pre-trained language model based on social relationship information such as friend relationships, guild relationships, and team history information in the digital twin information.

[0088] In practical applications, the representation of relationships can take the form of topological structures, knowledge graphs, etc., and this specification does not impose specific limitations. That is, the digital twin information of each target game player's digital twin can be input into a pre-trained model, allowing the pre-trained model to generate a social topological structure or social knowledge graph between the target digital twins based on social relationship information such as friend relationships, guild relationships, and team history information. Alternatively, the social topological structure or social knowledge graph between the target digital twins can be constructed manually. In this social topological structure or social knowledge graph, nodes represent target digital twins, and edges represent the strength of social relationships between target digital twins, such as intimacy values, interaction frequency, or guild hierarchy. The edges in the social topological structure or social knowledge graph are directed, and the direction of the edges represents information such as the direction of transmission of evaluation information for the content to be evaluated. For example, if digital twin A is the game group owner and digital twin B is a game group member, then the game group owner digital twin A will share the content to be evaluated and the evaluation information for that content with the group member digital twin B.

[0089] It should be noted that social communication simulation can be understood as utilizing the social topology or social knowledge graph between target digital twins to simulate, through an interactive reasoning process, the dynamic evolution of how the content to be evaluated is perceived, discussed, accepted, or resisted among the real game players corresponding to each target digital twin. Alternatively, social communication simulation can be understood as simulating, within a virtual social network (social topology or social knowledge graph) composed of target digital twins, the dynamic evolution of the formation of opinions, emotional contagion, and behavioral diffusion among the real game players corresponding to each target digital twin regarding the content to be evaluated.

[0090] In practical applications, a pre-trained language model can be used to simulate the social propagation of the content to be evaluated based on the relationships, the content to be evaluated, and the digital twin information of each target digital twin. The evaluation information corresponding to each digital twin after the social propagation simulation can then be obtained. Alternatively, a rule-based multi-agent simulation system, an improved SIR-type epidemic propagation model, an Opinion Dynamics Model (ODM), or a graph neural network-driven influence diffusion algorithm can be used to simulate the transmission, evolution, and convergence process of evaluation information for the content to be evaluated in a virtual social network. This simulates the social propagation of the content to be evaluated and obtains the evaluation information corresponding to each digital twin after the social propagation simulation. This specification does not impose specific restrictions on the methods used for social propagation simulation, as long as they can simulate the dynamically evolving group evaluation behavior under the influence of relationships, based on the relationships between the target digital twins, the content to be evaluated, and the digital twin information of each target digital twin.

[0091] By generating assessment information based on relationships, and taking into account the social relationships of the real users corresponding to the digital twins, the accuracy of the generated assessment information can be improved.

[0092] In one or more embodiments of this specification, social propagation simulation of the content to be evaluated is performed based on the aforementioned association, the content to be evaluated, and the digital twin information of each target digital twin, including: Based on the aforementioned association, a starting target digital twin and associated target digital twins are determined. Based on the digital twin information of the initial target digital twin and the content to be evaluated, evaluation information corresponding to the initial target digital twin is generated, and based on the evaluation information corresponding to the initial target digital twin, the digital twin information of the associated target digital twin, and the content to be evaluated, evaluation information corresponding to the associated target digital twin is generated. The associated target digital twin is used again as the starting target digital twin, and the associated target digital twins associated with the starting target digital twin are re-determined to continue generating evaluation information until the preset target conditions are met.

[0093] In practical applications, the starting and target digital twins can be determined based on the association relationships. The starting target digital twin can be understood as the target digital twin corresponding to any node in the social topology or social knowledge graph that has no upstream node. The associated target digital twin, related to the starting target digital twin, can be understood as the target digital twin corresponding to a downstream node in the social topology or social knowledge graph that is adjacent to the node corresponding to the starting target digital twin. For example... Figure 3 As shown, Figure 3 This is a schematic diagram of a starting target digital twin and an associated target digital twin provided as an embodiment of this specification.

[0094] It should be noted that the target digital twin often has a higher level of activity, gaming and social identity, etc., compared to the associated target digital twin.

[0095] The preset target conditions can be set based on actual needs and circumstances. These preset target conditions include, but are not limited to, the generation of evaluation information for all target digital twins with reachable paths in the social topology or social knowledge graph, and the achievement of preset propagation rounds, etc.

[0096] Based on the foregoing, the evaluation information in one or more embodiments of this specification may include first evaluation information corresponding to each target digital twin independently predicted, and second evaluation information corresponding to each target digital twin obtained through social communication simulation prediction under the influence of correlation. Furthermore, since the evaluation information may include sentiment text of the content to be evaluated, behavioral intention text of the content to be evaluated, reason text for taking that behavioral intention, and social expression text for the content to be evaluated, the first evaluation information may include first sentiment text of the content to be evaluated, first behavioral intention text of the content to be evaluated, first reason text for taking that behavioral intention, and first social expression text for the content to be evaluated; the second evaluation information may include second sentiment text of the content to be evaluated, second behavioral intention text of the content to be evaluated, second reason text for taking that behavioral intention, and second social expression text for the content to be evaluated.

[0097] By employing the aforementioned multi-round social simulation approach, the evaluation information obtained from each target digital twin can not only reflect the individual digital twin's own evaluation attitude but also incorporate the evolution of that attitude under social influence. This allows for a more realistic simulation of the dynamic evolution of game players in the real world, improving the accuracy of the generated evaluation information. Furthermore, it enables the comparison of the differences between the target digital twin's own evaluation attitude and the evaluation attitude after incorporating social influence, thereby analyzing the reasons for the changes, identifying potential public opinion risks, and analyzing the inflection points leading to such risks.

[0098] Step 206: Obtain the evaluation results for the content to be evaluated based on the evaluation information.

[0099] As mentioned earlier, the evaluation information may include first evaluation information corresponding to each target digital twin independently predicted, and second evaluation information corresponding to each target digital twin obtained by social communication simulation prediction under the influence of correlation.

[0100] In one or more embodiments of this specification, the first evaluation information and the second evaluation information corresponding to each target digital twin can be used as the evaluation result for the content to be evaluated.

[0101] In one or more embodiments of this specification, the first evaluation information and the second evaluation information can be compared to obtain a comparison result, which can reflect the difference between individual intuition and group evolution. The comparison result, the first evaluation information, and the second evaluation information corresponding to each target digital twin are then used as the evaluation result. In practical applications, the target digital twins whose evaluation information has changed and their number can be obtained based on the comparison result. For example, the first evaluation information corresponding to target digital twin C is: "This new game equipment is pretty good, I want to buy it." The second evaluation information corresponding to target digital twin C under the influence of its associated digital twins is: "Everyone thinks this new game equipment is bad, so I also think it's bad." It is evident that the evaluation attitude of target digital twin C has changed.

[0102] Furthermore, risk information can be generated based on the number of target digital twins whose assessment information has changed. Specifically, if the number of target digital twins whose assessment information has changed is greater than a first preset number, it can be determined that the current content to be assessed is at risk, and thus risk information representing the presence of risk in the current content to be assessed can be generated. If the number of target digital twins whose assessment information has changed is not greater than the first preset number, it can be determined that the current content to be assessed is not at risk, and thus risk information representing the absence of risk in the current content to be assessed can be generated.

[0103] Furthermore, the target digital twin whose assessment information has changed is used as a reference digital twin. Then, the number of reference digital twins among the associated target digital twins can be determined. Based on the number of reference digital twins corresponding to each target digital twin, the risk level of each target digital twin is determined. Specifically, the number of reference digital twins corresponding to a target digital twin can be positively correlated with the risk level of the target digital twin; that is, the more reference digital twins a target digital twin has, the higher its risk level, and the fewer reference digital twins a target digital twin has, the lower its risk level.

[0104] Therefore, in one or more embodiments of this specification, the comparison results, risk information, risk levels corresponding to each target digital twin, first assessment information, and second assessment information can be used as the assessment results. Alternatively, a risk analysis report can be generated based on the comparison results, risk information, risk levels corresponding to each target digital twin, first assessment information, and second assessment information, and this risk analysis report can also be used as part of the assessment results. Based on this risk analysis report, one can intuitively view the information of each target digital twin, the risk information, and the nodes and causes leading to the risk.

[0105] Among them, risk analysis reports can be generated based on pre-trained language models.

[0106] Furthermore, risk information can be generated based on the number of target digital twins for specified assessment information according to statistical assessment information. Specifically, assessment results are obtained for the content to be assessed based on each assessment piece of information, including: Determine the number of digital twins for key objectives, where the evaluation information corresponding to the digital twins of key objectives is the specified evaluation information; Based on the quantity, generate risk information for the content to be evaluated; The assessment results for the content to be assessed are obtained based on the assessment information and the risk information.

[0107] The specified evaluation information can be preset. The specified evaluation information can be evaluation information that indicates that the evaluation of the content to be evaluated is a positive evaluation, or it can be evaluation information that indicates that the evaluation of the content to be evaluated is a negative evaluation.

[0108] In this specification, the aforementioned risk information is referred to as first risk information, and the risk information generated based on the number of key target digital twins is referred to as second risk information. In one or more embodiments of this specification, when the number of key target digital twins is greater than a second preset number, it can be determined that the current content to be evaluated is at risk, and therefore second risk information representing the existence of risk in the current content to be evaluated can be generated. When the number of target digital twins whose evaluation information has changed is not greater than a second preset number, it can be determined that the current content to be evaluated is not at risk, and therefore second risk information representing the absence of risk in the current content to be evaluated can be generated.

[0109] It should be noted that when there is a conflict between the first and second risk information, manual review or verification based on a pre-trained language model can be performed to determine the existence of abnormal risk information.

[0110] In one or more embodiments of this specification, obtaining an evaluation result for the content to be evaluated based on various evaluation information includes: If it is determined that the content to be evaluated is at risk based on the risk information, target keywords are extracted based on each evaluation information, wherein the target keywords are keywords that appear more than a specified number of times in each evaluation information; Based on the target keywords and various assessment information, generate an analysis report on the risks involved; The assessment results for the content to be assessed are obtained based on the assessment information, the risk information, and the analysis report.

[0111] In practical applications, when the risk information indicates that the content to be assessed is at risk, keyword mining and risk attribution analysis can be performed. Specifically, when the risk information indicates that the content to be assessed is at risk, target keywords can be extracted from each assessment information, wherein the target keywords are keywords that appear more than a specified number of times in each assessment information.

[0112] It should be noted that the specified number of times can be preset, and this manual does not impose specific restrictions.

[0113] In one or more embodiments of this specification, each evaluation piece of information can be segmented using a pre-trained language model, such as a large language model, to obtain segmented words, and target keywords that appear more than a specified number of times can be identified. Furthermore, the large language model can analyze the reasons for the risks posed by the content to be evaluated based on the target keywords and information from each target digital twin to obtain an analysis report. This allows the evaluation results for the content to be evaluated to be derived from the evaluation information, the risk information, and the analysis report, providing reliable information for the optimization of the target game, communication strategies, or emergency intervention.

[0114] Based on the above data processing methods, a digital twin can be constructed based on the user history data of real users and used to evaluate the content to be evaluated. Since the digital twin is constructed based on the user history data of real users, it is a mapping of real users. Therefore, by using the digital twin to evaluate the content to be evaluated in advance, the situation of real users not participating in the evaluation of the content to be evaluated can be avoided, and accurate and comprehensive evaluation results can be obtained.

[0115] Furthermore, by simulating social communication through digital twins constructed based on multiple data dimensions and corresponding virtual social networks, it is possible to predict in advance the reactions of game update content to be evaluated within the game player community. This can solve the problems of lag, bias, and inaccuracy in current evaluation methods, thereby improving the optimization strategy for game update content to be evaluated and enhancing the user experience.

[0116] Corresponding to the above method embodiments, this application also provides a data processing apparatus embodiment. Figure 4 A schematic diagram of the structure of a data processing apparatus according to one embodiment of this specification is shown. Figure 4 As shown, the device includes: The acquisition module 402 is configured to acquire the content to be evaluated and multiple target digital twins, wherein the target digital twins are pre-built based on the user history data of the target real users; The generation module 404 is configured to generate evaluation information for each target digital twin to evaluate the content to be evaluated, based on the content to be evaluated and the digital twin information of each target digital twin. The module 406 is configured to obtain the evaluation result for the content to be evaluated based on the evaluation information.

[0117] Optionally, the acquisition module 402 is further configured to select multiple target digital twins from each digital twin in the digital twin library according to a preset sampling strategy, wherein the digital twins in the digital twin library are constructed based on the user history data of real users.

[0118] Optionally, the acquisition module 402 is further configured to acquire user history data of multiple real users; generate digital twin information corresponding to each real user based on multiple preset data dimensions and user history data of each real user; and obtain digital twins corresponding to each real user based on the digital twin information corresponding to each real user.

[0119] Optionally, the acquisition module 402 is further configured to: determine first information based on user historical data in the interaction data dimension, wherein the first information represents the emotional information and / or social information of a real user; or determine second information based on user historical data in the behavior data dimension, wherein the second information represents the activity information and / or psychological information of a real user; or determine third information based on user historical data in the transaction data dimension, wherein the third information represents the transaction information of a real user; or determine fourth information based on user historical data in the attribute information dimension, wherein the fourth information represents the identity information of a real user; and obtain digital twin information based on at least one of the first information, the second information, the third information, and the fourth information.

[0120] Optionally, the generation module 404 is further configured to input the content to be evaluated and the digital twin information of the target digital twin to be processed into a pre-trained language model, so that the pre-trained language model simulates the evaluation process of the target digital twin on the content to be evaluated based on the digital twin information of the target digital twin to be processed, and obtains the evaluation information corresponding to the target digital twin to be processed, wherein the target digital twin to be processed is any one of the target digital twins.

[0121] Optionally, the generation module 404 is further configured to parse the content to be evaluated according to a preset parsing dimension, generate the evaluation dimension information corresponding to the content to be evaluated, and input the evaluation dimension information and the digital twin information of the target digital twin to be processed into a pre-trained language model.

[0122] Optionally, the generation module 404 is further configured to: obtain the association relationship between each target digital twin; and, based on the association relationship, the content to be evaluated, and the digital twin information of each target digital twin, perform social communication simulation for the content to be evaluated to obtain evaluation information of each target digital twin evaluating the content to be evaluated.

[0123] Optionally, the generation module 404 is further configured to: determine a starting target digital twin and associated target digital twins based on the association relationship; generate evaluation information corresponding to the starting target digital twin based on the digital twin information of the starting target digital twin and the content to be evaluated; generate evaluation information corresponding to the associated target digital twin based on the evaluation information corresponding to the starting target digital twin, the digital twin information of the associated target digital twin, and the content to be evaluated; re-use the associated target digital twin as the starting target digital twin, and re-determine the associated target digital twins associated with the starting target digital twin to continue generating evaluation information until the preset target conditions are met.

[0124] Optionally, the obtaining module 406 is further configured to: determine the number of key target digital twins, wherein the evaluation information corresponding to the key target digital twins is specified evaluation information; generate risk information for the content to be evaluated based on the number; and obtain an evaluation result for the content to be evaluated based on each evaluation information and the risk information.

[0125] Optionally, the obtaining module 406 is further configured to, when it is determined that the content to be evaluated has a risk based on the risk information, extract target keywords based on each evaluation information, wherein the target keywords are keywords that appear more than a specified number of times in each evaluation information; generate an analysis report that leads to the risk based on the target keywords and each evaluation information; and obtain an evaluation result for the content to be evaluated based on each evaluation information, the risk information, and the analysis report.

[0126] Based on the aforementioned data processing device, a digital twin can be constructed based on the user history data of real users and used to evaluate the content to be evaluated. Since the digital twin is constructed based on the user history data of real users, it is a mapping of real users. Therefore, by using the digital twin to evaluate the content to be evaluated in advance, the situation where real users do not participate in the evaluation of the content to be evaluated can be avoided, and accurate and comprehensive evaluation results can be obtained.

[0127] The above is an illustrative scheme of a data processing apparatus according to this embodiment. It should be noted that the technical solution of this data processing apparatus and the technical solution of the data processing method described above belong to the same concept. For details not described in detail in the technical solution of the data processing apparatus, please refer to the description of the technical solution of the data processing method described above.

[0128] Figure 5 A structural block diagram of a computing device 500 according to one embodiment of this specification is shown. The components of the computing device 500 include, but are not limited to, a memory 510 and a processor 520. The processor 520 is connected to the memory 510 via a bus 530, and a database 550 is used to store data.

[0129] The computing device 500 also includes an access device 540, which enables the computing device 500 to communicate via one or more networks 560. Examples of these networks include Public Switched Telephone Network (PSTN), Local Area Network (LAN), Wide Area Network (WAN), Personal Area Network (PAN), or combinations of communication networks such as the Internet. The access device 540 may include one or more of any type of wired or wireless network interface (e.g., a network interface card (NIC)), such as an IEEE 802.11 Wireless Local Area Network (WLAN) wireless interface, a Wi-MAX (Worldwide Interoperability for Microwave Access) interface, an Ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a Bluetooth interface, a Near Field Communication (NFC) interface, and so on.

[0130] In one embodiment of this application, the aforementioned components of the computing device 500 and Figure 5 Other components, not shown, can also be connected to each other, for example, via a bus. It should be understood that... Figure 5 The block diagram of the computing device shown is for illustrative purposes only and is not intended to limit the scope of this application. Those skilled in the art can add or replace other components as needed.

[0131] The computing device 500 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or personal computers (PCs). The computing device 500 can also be a mobile or stationary server.

[0132] The processor 520 is used to execute the following computer program / instruction, which, when executed by the processor, implements the steps of the above-mentioned digital twin construction method and data processing method.

[0133] The above is an illustrative scheme of a computing device according to this embodiment. It should be noted that the technical solution of this computing device belongs to the same concept as the technical solutions of the digital twin construction method and data processing method described above. For details not described in detail in the technical solution of the computing device, please refer to the description of the technical solutions of the digital twin construction method and data processing method described above.

[0134] An embodiment of this specification also provides a computer-readable storage medium storing a computer program / instructions that, when executed by a processor, implement the steps of the above-described digital twin construction method and data processing method.

[0135] The above is an illustrative scheme of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium belongs to the same concept as the technical solutions of the digital twin construction method and data processing method described above. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solutions of the digital twin construction method and data processing method described above.

[0136] An embodiment of this specification also provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the above-described digital twin construction method and data processing method.

[0137] The above is an illustrative scheme of a computer program product according to this embodiment. It should be noted that the technical solution of this computer program product belongs to the same concept as the technical solutions of the above-described digital twin construction method and data processing method. For details not described in detail in the technical solution of the computer program product, please refer to the description of the technical solutions of the above-described digital twin construction method and data processing method.

[0138] The foregoing has described specific embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0139] The computer instructions include computer program code, which may be in the form of source code, object code, executable file, or certain intermediate forms. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium may be appropriately added or removed according to the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0140] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0141] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0142] The preferred embodiments disclosed above are merely illustrative of this application. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this application. These embodiments are selected and specifically described in this application to better explain the principles and practical applications of this application, thereby enabling those skilled in the art to better understand and utilize this application. This application is limited only by the claims and their full scope and equivalents.

Claims

1. A data processing method, characterized in that, include: Acquire the content to be evaluated and multiple target digital twins, wherein the target digital twins are pre-built based on the user history data of the target real users; Based on the content to be evaluated and the digital twin information of each target digital twin, evaluation information is generated for each target digital twin to evaluate the content to be evaluated. The evaluation results for the content to be evaluated are obtained based on the evaluation information.

2. The method as described in claim 1, characterized in that, Obtain multiple target digital twins, including: Multiple target digital twins are selected from each digital twin in the digital twin library according to a preset sampling strategy. The digital twins in the digital twin library are constructed based on the user history data of real users.

3. The method as described in claim 1 or 2, characterized in that, Digital twins are constructed in the following ways: Obtain user history data from multiple real users; Based on multiple preset data dimensions and the user history data of each real user, generate digital twin information corresponding to each real user; Based on the digital twin information corresponding to each real user, obtain the digital twin corresponding to each real user.

4. The method as described in claim 3, characterized in that, Based on multiple preset data dimensions and the user history data of each real user, digital twin information corresponding to each real user is generated, including: The first piece of information is determined based on user history data from the interaction data dimension, wherein the first piece of information represents the emotional information and / or social information of real users; or, The second information is determined based on historical user data from the behavioral data dimension, wherein the second information represents the activity information and / or psychological information of real users; or, The third piece of information is determined based on historical user data from the transaction data dimension, wherein the third piece of information represents the transaction information of a real user; or, The fourth information is determined based on the user's historical data in the attribute information dimension, wherein the fourth information represents the identity information of the real user; Digital twin information is obtained based on at least one of the first information, the second information, the third information, and the fourth information.

5. The method as described in claim 1, characterized in that, Based on the content to be evaluated and the digital twin information of each target digital twin, evaluation information is generated for each target digital twin to evaluate the content to be evaluated, including: The content to be evaluated and the digital twin information of the target digital twin are input into a pre-trained language model, so that the pre-trained language model simulates the evaluation process of the target digital twin on the content to be evaluated based on the digital twin information of the target digital twin, and obtains the evaluation information corresponding to the target digital twin, wherein the target digital twin is any one of the target digital twins.

6. The method as described in claim 5, characterized in that, The evaluation content and the digital twin information of the target digital twin are input into a pre-trained language model, including: The content to be evaluated is analyzed according to the preset analysis dimensions, and the corresponding evaluation dimension information of the content to be evaluated is generated. The dimensional information to be evaluated and the digital twin information of the target digital twin to be processed are input into the pre-trained language model.

7. The method as described in claim 1, characterized in that, Based on the content to be evaluated and the digital twin information of each target digital twin, evaluation information is generated for each target digital twin to evaluate the content to be evaluated, including: Obtain the relationships between the target digital twins; Based on the aforementioned relationships, the content to be evaluated, and the digital twin information of each target digital twin, a social communication simulation is performed on the content to be evaluated to obtain evaluation information of each target digital twin evaluating the content to be evaluated.

8. The method as described in claim 7, characterized in that, Based on the aforementioned relationships, the content to be evaluated, and the digital twin information of each target digital twin, a social media communication simulation targeting the content to be evaluated is performed, including: Based on the aforementioned association, a starting target digital twin and associated target digital twins are determined. Based on the digital twin information of the initial target digital twin and the content to be evaluated, evaluation information corresponding to the initial target digital twin is generated, and based on the evaluation information corresponding to the initial target digital twin, the digital twin information of the associated target digital twin, and the content to be evaluated, evaluation information corresponding to the associated target digital twin is generated. The associated target digital twin is used again as the starting target digital twin, and the associated target digital twins associated with the starting target digital twin are re-determined to continue generating evaluation information until the preset target conditions are met.

9. The method as described in claim 1, characterized in that, The evaluation results for the content to be evaluated are obtained based on the evaluation information, including: Determine the number of digital twins for key objectives, where the evaluation information corresponding to the digital twins of key objectives is the specified evaluation information; Based on the quantity, generate risk information for the content to be evaluated; The assessment results for the content to be assessed are obtained based on the assessment information and the risk information.

10. The method as described in claim 9, characterized in that, The evaluation results for the content to be evaluated are obtained based on the evaluation information, including: If it is determined that the content to be evaluated is at risk based on the risk information, target keywords are extracted based on each evaluation information, wherein the target keywords are keywords that appear more than a specified number of times in each evaluation information; Based on the target keywords and various assessment information, generate an analysis report on the risks involved; The assessment results for the content to be assessed are obtained based on the assessment information, the risk information, and the analysis report.

11. A method for constructing a digital twin, characterized in that, include: Obtain user history data from multiple real users; Based on multiple preset data dimensions and the user history data of each real user, generate digital twin information corresponding to each real user; Based on the digital twin information corresponding to each real user, obtain the digital twin corresponding to each real user.

12. A data processing apparatus, characterized in that, include: The acquisition module is configured to acquire the content to be evaluated and multiple target digital twins, wherein the target digital twins are pre-built based on the user history data of the target real users; The generation module is configured to generate evaluation information for each target digital twin to evaluate the content to be evaluated, based on the content to be evaluated and the digital twin information of each target digital twin. The acquisition module is configured to obtain the evaluation results for the content to be evaluated based on the evaluation information.

13. A computing device, characterized in that, include: Memory and processor; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions, which, when executed by the processor, implement the steps of the method according to any one of claims 1 to 10.

14. A computer-readable storage medium storing a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 10.

15. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 10.

Images