A data space data resource supply and demand matching method based on large model

Abstract

The present application discloses a data space data resource supply and demand matching method, device, equipment and storage medium based on a large model, including: in response to an input operation of a demand description statement for a target digital object, determining the demand characteristics of the target digital object according to the direct description features extracted from the demand description statement and the potential description features obtained by reasoning operations from the direct description features; searching in a digital tag set according to the demand characteristics to obtain a target digital tag matching the demand characteristics from the digital tag set; determining a digital object matching the target digital tag as the target digital object. According to the method provided by the present application, the accuracy and real-time performance of the matching are improved, and the problem of insufficient efficiency and accuracy of data matching in the data space in the prior art is solved.

Description

Technical Field

[0001] The present application belongs to the field of data retrieval, and specifically relates to a method, device, equipment and storage medium for matching supply and demand of data space data resources based on a large model. Background Art

[0002] With the rapid development of big data and artificial intelligence technologies, all industries have put forward urgent demands for efficient data supply and demand matching. Especially in the data space environment, hundreds of millions of digital objects need to be matched efficiently and accurately to maximize the use and innovation of data value.

[0003] At present, Digital Object Architecture (DOA) is one of the mainstream technologies for data resource management and exchange. As the core element, digital objects abstract data resources through three parts: identification, metadata, and entity. Existing supply and demand matching systems mainly rely on users to manually search directories or perform simple keyword matching to retrieve digital objects. These systems use the Digital Object Interface Protocol (DOIP) and the Digital Object Identifier Resolution Protocol (DO-IRP) to regulate data access and operations.

[0004] Although these technologies have achieved certain results in data management and basic retrieval, the existing technologies rely on manual search and simple keyword matching by users, which makes them inefficient and difficult to fully mine when dealing with massive and complex data requirements, resulting in poor user experience. Summary of the invention

[0005] The present application aims to provide a method, device, equipment and storage medium for matching supply and demand of data resources in a data space based on a large model, so as to at least solve the problem of insufficient efficiency and accuracy of data matching in the data space.

[0006] In a first aspect, the present application discloses a method for matching supply and demand of data space data resources based on a large model, including:

[0007] In response to an input operation of a requirement description statement of a target digital object, determining the requirement features of the target digital object according to direct description features extracted from the requirement description statement and potential description features obtained by inference operation from the direct description features;

[0008] According to the requirement characteristics, searching in the digital tag set to obtain a target digital tag matching the requirement characteristics from the digital tag set; each of the digital tags has a corresponding digital object;

[0009] A digital object matching the target digital tag is determined as the target digital object.

[0010] In a second aspect, the embodiment of the present application further discloses a data space data resource supply and demand matching device based on a large model, comprising:

[0011] A feature generation module, configured to determine the requirement features of the target digital object in response to an input operation of a requirement description statement of the target digital object, based on direct description features extracted from the requirement description statement and potential description features obtained through inference operations from the direct description features;

[0012] A tag retrieval module, used to search in a digital tag set according to the demand characteristics, so as to obtain a target digital tag matching the demand characteristics from the digital tag set; each of the digital tags has a corresponding digital object;

[0013] The digital object module is used to determine the digital object matching the target digital tag as the target digital object.

[0014] In the third aspect, an embodiment of the present application further discloses an electronic device, comprising a processor and a memory, wherein the memory stores programs or instructions that can be run on the processor, and when the programs or instructions are executed by the processor, the steps of the data space data resource supply and demand matching method based on a large model as described in the first aspect are implemented.

[0015] In a fourth aspect, an embodiment of the present application further discloses a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the data space data resource supply and demand matching method based on a large model as described in the first aspect are implemented.

[0016] In summary, in the embodiments of the present application, by generating demand features, not only based on explicit descriptions, but also based on direct descriptions, potential demands can be inferred, thereby improving the comprehensiveness and accuracy of demand understanding; then, based on the demand features, a search is performed in the digital tag set to obtain a target digital tag that matches the demand features, and the demand features are used to search in the digital tag set, so that the search results are more accurate and relevant. By utilizing the use of digital tags, digital objects can be better characterized, and fast and accurate matching can be achieved, thereby overcoming the inefficiency of traditional manual search and simple keyword matching; finally, the target digital object is determined by matching the target digital tag, which can quickly lock the most relevant data object, improve matching efficiency and accuracy, and make full use of data resources. Therefore, based on the method of the embodiment of the present application, the demand features and digital tags are generated by a large model for centralized retrieval, which reduces the time for users to manually search and automates the complex data matching process; and utilizes the inference and intelligent matching mechanism of potential descriptive data to enable the system to mine and utilize more high-value data resources and avoid missing relevant data. Ultimately, the entire data matching process is automatically executed by the system, and users do not need to intervene frequently, which simplifies operations and improves the accuracy and real-time nature of matching, thereby enhancing user experience and solving the problem of insufficient efficiency and accuracy of data matching in the data space in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In the attached picture:

[0018] Figure 1 It is a flowchart of the steps of a method for matching supply and demand of data space data resources based on a large model provided in an embodiment of the present application;

[0019] Figure 2 It is a flowchart of the steps of another method for matching supply and demand of data space data resources based on a large model provided in an embodiment of the present application;

[0020] Figure 3 It is a program execution logic diagram under the method provided in the embodiment of the present application;

[0021] Figure 4 It is a block diagram of a data space data resource supply and demand matching device based on a large model provided in an embodiment of the present application;

[0022] Figure 5 is a block diagram of an electronic device according to an embodiment of the present application;

[0023] Figure 6 It is a block diagram of an electronic device of another embodiment provided by the embodiments of the present application. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] The terms "first", "second", etc. in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0026] Figure 1 This embodiment provides a data space data resource supply and demand matching method based on a large model.

[0027] The method may include the following steps:

[0028] Step 101, in response to an input operation of a requirement description statement for a target digital object, the requirement features of the target digital object are determined based on direct description features extracted from the requirement description statement and potential description features obtained through inference operations from the direct description features.

[0029] In some embodiments of the present application, the demand features of the target digital object will be generated in response to the user's demand description statement for the target digital object. The demand features include two parts: one part is the direct description features of the target digital object, that is, the needs clearly expressed by the user; the other part is the potential description features obtained by reasoning based on the direct description data, that is, when the system analyzes the user's needs, it infers the additional information that the user may need based on the existing information. In this way, the system can understand the user's needs more comprehensively and improve the accuracy and relevance of matching.

[0030] For example, a user enters a demand description sentence in the system, "need the latest research materials on big data analysis technology". The system first extracts the direct descriptive data clearly expressed by the user, that is, "the latest research materials on big data analysis technology". Then, based on this descriptive data, the system infers that the user may also need some related potential descriptive data, such as "research materials containing practical application cases" or "cutting-edge technology trends related to big data analysis". By generating these demand features, the system can understand the user's needs more accurately, thereby providing more matching results in the subsequent retrieval process.

[0031] Step 102: searching in the digital tag set according to the demand characteristics to obtain a target digital tag matching the demand characteristics from the digital tag set.

[0032] Each digital tag has a corresponding digital object.

[0033] In some embodiments of the present application, a search is performed in a digital tag set based on the generated demand characteristics. The digital tag set contains multiple digital tags, each of which corresponds to a specific digital object. The purpose of the search is to find a target digital tag that matches the demand characteristics from these digital tags. In this way, the system can quickly locate the most relevant target digital tag from a large number of digital tags and ensure that the corresponding digital object meets the user's needs, thereby improving the accuracy and efficiency of the search results.

[0034] For example, the requirements input by the user include "the latest research results of machine learning algorithms" and "technical documents containing application examples". The system searches the digital tag set to find digital tags that match these requirements, such as "2023 machine learning algorithm review" and "machine learning application case study". The digital objects corresponding to these digital tags may be related research papers, technical reports, etc. By matching these digital tags, the system can quickly lock the target digital objects that meet the user's needs, improving the retrieval efficiency and the accuracy of the results.

[0035] Step 103: determine the digital object matching the target digital tag as the target digital object.

[0036] In some embodiments of the present application, digital objects matching the target digital tags are determined as target digital objects. In the previous step, the system has found the target digital tags matching the required characteristics in the digital tag set. Next, the system will locate the digital objects corresponding to them through these target digital tags. Determining these digital objects matching the target digital tags as target digital objects means that the system has successfully found specific data objects that meet the user's needs. This step ensures the accuracy of the retrieval results and enables users to obtain the most relevant data resources.

[0037] For example, the user's demand feature is "the latest research on natural language processing technology", and the system finds the target digital tag "2023 Natural Language Processing Technology Review" that matches this demand feature in the digital tag set. The digital objects corresponding to these digital tags may be related research papers or technical reports. The system identifies these digital objects as target digital objects and provides them to the user to meet the user's needs. In this way, the system can efficiently and accurately match the digital objects required by the user, improve the utilization rate of data resources and user satisfaction.

[0038] In summary, in the embodiments of the present application, by generating demand features, not only based on explicit descriptions, but also based on direct descriptions, potential demands can be inferred, thereby improving the comprehensiveness and accuracy of demand understanding; then, based on the demand features, a search is performed in the digital tag set to obtain a target digital tag that matches the demand features, and the demand features are used to search in the digital tag set, so that the search results are more accurate and relevant. By utilizing the use of digital tags, digital objects can be better characterized, and fast and accurate matching can be achieved, thereby overcoming the inefficiency of traditional manual search and simple keyword matching; finally, the target digital object is determined by matching the target digital tag, which can quickly lock the most relevant data object, improve matching efficiency and accuracy, and make full use of data resources. Therefore, based on the method of the embodiment of the present application, the demand features and digital tags are generated by a large model for centralized retrieval, which reduces the time for users to manually search and automates the complex data matching process; and utilizes the inference and intelligent matching mechanism of potential descriptive data to enable the system to mine and utilize more high-value data resources and avoid missing relevant data. Ultimately, the entire data matching process is automatically executed by the system, and users do not need to intervene frequently, which simplifies operations and improves the accuracy and real-time nature of matching, thereby enhancing user experience and solving the problem of insufficient efficiency and accuracy of data matching in the data space in the prior art.

[0039] Figure 2 This is another data space data resource supply and demand matching method based on a large model provided in an embodiment of the present application.

[0040] The method may include the following steps:

[0041] Step 201, determining description data of a digital object.

[0042] In some embodiments of the present application, the description data of the digital object will be determined first. The description data is a detailed description and feature description of the digital object, which includes not only the basic information of the digital object, but also various attributes related to the digital object. These description data are the basis for the system to generate digital tags and perform retrieval in subsequent steps. By clarifying the description data, the system can more accurately understand the content and characteristics of the digital object, thereby providing more accurate results in the user demand matching process.

[0043] For example, a digital object is a research paper on "Application of Machine Learning in Medical Diagnosis". In step 201, the system extracts the descriptive data of the research paper, including the title, author, publication date, abstract, keywords, main content and other information of the paper. These descriptive data describe the various features and attributes of the research paper in detail, laying the foundation for the generation and retrieval of digital tags in the subsequent steps. In this way, the system can better understand the digital object, thereby improving the accuracy of subsequent matching and retrieval.

[0044] Optionally, step 201 may process different types of digital objects according to the following sub-steps:

[0045] Sub-step 2011, when the digital object is text data, text recognition is performed on the digital object, so as to use the knowledge graph of the digital object obtained through text recognition as description data.

[0046] In some embodiments of the present application, text recognition is performed on the text data for the case where the digital object is text data. The purpose of text recognition is to extract meaningful information and structured data from the text data, which can be organized into a knowledge graph as descriptive data of the digital object. A knowledge graph is a data structure used to represent entities and their relationships. By converting text data into a knowledge graph, the system can more fully understand the content and associations of the text data, thereby improving accuracy and relevance in subsequent retrieval and matching processes.

[0047] For example, a digital object is a research paper on "the application of artificial intelligence in medical diagnosis". The system first performs text recognition on the text content of the research paper and extracts key information, such as "artificial intelligence", "medical diagnosis", "application examples", etc. Then, the system uses this extracted information to build a knowledge graph, representing the relationship between "artificial intelligence" and "medical diagnosis" and the specific "application examples" as nodes and edges. This knowledge graph, as descriptive data, enables the system to better understand the content of the research paper, thereby providing more accurate results when users search for related topics.

[0048] Sub-step 2012: When the digital object is unstructured data, content recognition is performed on the digital object, so as to use the description text of the digital object obtained through content recognition as description data.

[0049] In some embodiments of the present application, the digital objects of unstructured data are targeted, involving content recognition of these data. Unstructured data refers to data that has not been sorted and classified, such as pictures, videos, audio or free text. The purpose of content recognition is to convert unstructured data into structured information so that it can be understood and used by the system. Through content recognition, the system can extract useful descriptive text from unstructured data and use it as descriptive data of digital objects, thereby improving the accuracy of data processing and retrieval.

[0050] For example, a digital object is a video about "smart home devices". At this point, the system can first perform content recognition on the video and extract key information, such as dialogue, subtitles, and scene descriptions in the video. The system converts this information into descriptive text, such as "the video content includes an introduction to smart home devices, functional demonstrations, and user experience". These descriptive texts serve as descriptive data for digital objects, enabling the system to better understand the content of the video, thereby providing more accurate results when users search for related topics.

[0051] Sub-step 2013, when the digital object is multimodal data, content recognition is performed on sub-data in different modes in the digital object respectively, so as to obtain multiple description sub-texts of the digital object through content recognition, and description data is generated according to the obtained multiple description sub-texts.

[0052] In some embodiments of the present application, for digital objects of multimodal data, content recognition is performed on sub-data in different modes. Multimodal data refers to objects containing multiple data forms, such as data containing text, images, audio and video at the same time. By performing content recognition on sub-data in different modes, corresponding description texts can be extracted from each mode. These descriptive sub-texts are specific descriptions of the content of each modal data. By combining these descriptive sub-texts to generate complete description data, the system can fully understand all aspects of multimodal data and improve the accuracy of data processing and retrieval.

[0053] For example, a digital object is a report on "smart city technology" that contains text, pictures, and videos. Then the system can perform text recognition on the text part to extract the title and main content of the report; perform image recognition on the picture part to extract key information in the image, such as the city layout map and technical schematic diagram; perform video content recognition on the video part to extract the explanation content and demonstration operations in the video. These identified descriptive sub-texts include "Report overview of smart city technology", "Pictures showing city layout map", "Video demonstration of technical operation process", etc. By combining these descriptive sub-texts, the system generates complete description data, enabling it to provide more accurate and comprehensive results when users search for relevant content.

[0054] Optionally, in order to generate description data according to the obtained multiple description subtexts, sub-step 2013 includes the following sub-steps:

[0055] Sub-step 20131, inputting multiple description sub-texts into the question guessing model respectively to obtain question sub-texts for the digital object.

[0056] The question sub-text is a description statement of the demand for the digital object inferred from the description sub-text.

[0057] In some embodiments of the present application, multiple description subtexts are input into the question guessing model to obtain question subtexts for the digital object. The question guessing model is a tool for generating potential demand descriptions, which infers the questions that the user may ask about the digital object based on the input description subtext. These question subtexts are further abstractions and inferences of the description subtexts, and by asking questions, the potential demand for the digital object is more comprehensively understood and characterized. In this way, the system can better predict and meet the needs of users.

[0058] For example, a digital object is a multimodal report on "The Application of Blockchain Technology in Supply Chain Management", which contains text, pictures and videos. The system identifies these contents and generates descriptive subtexts, such as "The advantages of blockchain in the supply chain", "Application examples in supply chain management", etc. Then the system can input these descriptive subtexts into the question guessing model, and the model generates question subtexts, such as "How does blockchain improve the transparency of the supply chain?", "What are the specific application scenarios in supply chain management?". These question subtexts help the system to have a deeper understanding of the user's possible needs for digital objects, thereby providing more accurate results in the subsequent retrieval and matching process.

[0059] Sub-step 20132, using the description sub-text and / or the question sub-text as description data.

[0060] In some embodiments of the present application, description subtext and / or question subtext are used as description data. The description subtext is a specific description of the digital object, while the question subtext is a question about possible needs inferred from the description subtext. Using these two types of text as description data can more comprehensively characterize the content and potential uses of the digital object, thereby improving the relevance and accuracy of the data in subsequent retrieval and matching processes. By combining the description subtext and the question subtext, the system can better meet the diverse needs of users.

[0061] For example, a digital object is a multimodal report on "The Application of 5G Technology in Smart Cities". The description subtexts generated by the system include "Overview of 5G Technology", "Smart City Application Cases", "Technical Challenges and Future Development", etc. At the same time, the system generates question subtexts through the question guessing model, such as "How does 5G technology improve the security of smart cities?", "Which cities have successfully applied 5G technology?", etc. Then the system can use all or part of these description subtexts and question subtexts as description data, so that when users retrieve relevant information, the system can provide more accurate and comprehensive results, improve data utilization and user satisfaction.

[0062] It should be emphasized that in some specific operation process embodiments of the present application, if only the description sub-text is used as the description data, the sub-step 20131 can be omitted at this time, that is, the causal relationship between sub-step 20132 and sub-step 20131 is only necessary when the question sub-text needs to be used as all or part of the description data.

[0063] Step 202: Generate a digital tag of the digital object according to the description data of the digital object, and store the digital object in a database according to the generated digital tag.

[0064] In some embodiments of the present application, digital tags of digital objects are generated based on the description data of the digital objects, and the digital objects are stored in a database according to these generated digital tags. Digital tags are an index and identification of digital objects, so that these objects can be effectively retrieved and managed in the database. By generating digital tags for each digital object, the system can classify and identify digital objects, thereby improving the accessibility and retrieval efficiency of data. This step not only enables digital objects to be efficiently managed by the system, but also provides a basis for subsequent retrieval and matching.

[0065] For example, the system determines that a certain digital object is a white paper on "the application of blockchain technology in the financial field". In step 202, the system generates corresponding digital tags, such as "blockchain technology", "financial application", "white paper", "2024", etc., based on the description data of the white paper. These digital tags represent the core content and characteristics of the white paper. Then, the system stores the white paper in the database based on these generated digital tags, so that it can be quickly located and accessed when the user retrieves relevant content. In this way, the system can not only effectively manage digital objects, but also improve the utilization efficiency of data resources.

[0066] Optionally, step 202 includes the following sub-steps:

[0067] Sub-step 2021, taking each description data of the digital object as a digital label of the digital object.

[0068] In some embodiments of the present application, each description data of a digital object is used as a digital tag of the digital object. The description data is a detailed description of the digital object, including various attributes and characteristics. By using these description data as digital tags, the system can generate multiple digital tags for each digital object. These digital tags are used to identify and index digital objects so that they can be efficiently retrieved and managed in the database. In this way, the system can classify and identify digital objects according to different description data, thereby improving the accessibility and retrieval efficiency of the data.

[0069] For example, a digital object is a research report on "Application of Internet of Things Technology in Smart Agriculture". The system extracts the descriptive data of the report, including "Internet of Things Technology", "Smart Agriculture", "Application Examples", "Water-saving Irrigation Technology", etc. Then the system can use these descriptive data as digital tags to generate multiple digital tags for the research report. In this way, when the user searches for "smart agriculture" or "water-saving irrigation technology" in the database, the research report can be quickly located and retrieved. In this way, the system can manage and retrieve digital objects more accurately, improving data utilization efficiency and user experience.

[0070] Sub-step 2022: storing all digital objects in a database, and establishing a correspondence between digital tags and digital objects in the database based on all digital tags.

[0071] In some embodiments of the present application, all digital objects are stored in a database, and a correspondence between the digital tag and the digital object is established based on the digital tag of each digital object. Storing all digital objects in a database ensures centralized management and efficient access to data. Establishing a correspondence based on the digital tag, that is, an index between the digital tag and the digital object, enables the system to quickly locate and retrieve digital objects associated with a certain digital tag. This step improves the speed and accuracy of data retrieval by establishing a clear mapping relationship between tags and objects in the database.

[0072] For example, when the system processes multiple reports on "intelligent transportation system", it stores each report in the database. The descriptive data of these reports include "intelligent transportation", "traffic management technology", "Internet of Vehicles", etc. Then the system can generate multiple digital tags for each report and store these reports in the database. At the same time, the system establishes a corresponding relationship between digital tags and these reports in the database. For example, when a user searches for "Internet of Vehicles", the system can quickly locate relevant reports through tags, thereby providing efficient retrieval results. This approach enables the system to better manage and utilize digital objects, improve the utilization efficiency of data resources and improve user experience.

[0073] It should be emphasized that in the embodiments of the present application, there is no further restriction on the type and requirements of the database. For example, in some implementations of the present application, a vector database, such as Milvus or Chroma, or a structured data storage database, such as neo4j, or other databases with comprehensive performance, such as ElasticSearch, can be selected. In some other embodiments of the present application, the data can also be directly stored in other Internet databases. Those skilled in the art can make a choice based on the actual requirements of the technical solution.

[0074] Step 203, in response to the input operation of the requirement description statement of the target digital object, the requirement features of the target digital object are determined based on the direct description features extracted from the requirement description statement and the potential description features obtained by inference operation from the direct description features.

[0075] The method shown in this step has been explained in step 101 and will not be repeated here.

[0076] Optionally, step 203 includes the following sub-steps:

[0077] Sub-step 2031, in response to the input operation of the requirement description statement, generating direct description features of the target digital object.

[0078] The direct description feature is used to represent the direct description data of the target digital object.

[0079] In some embodiments of the present application, a direct description feature of the target digital object is generated in response to the demand description statement. The direct description feature is a specific manifestation of the user's explicit demand and description data for the target digital object. The purpose of this step is to extract key information from the user's demand description and convert it into a direct description feature so that demand analysis and matching can be performed more accurately in subsequent steps.

[0080] For example, the user inputs a requirement description statement of "need the latest literature on machine learning algorithms". Then the system can extract key elements in the requirement, such as "latest", "machine learning algorithm", "literature", etc., and generate direct description features from these elements. These direct description features specifically represent the user's explicit needs for digital objects, and serve as the basis for subsequent inference of potential description features and generation of demand features. In this way, the system can accurately understand and record the user's direct demand information.

[0081] Sub-step 2032, inputting the direct description features into the first proxy model to perform reasoning operations through the thought chain engineering executed by the first proxy model to obtain the potential description features of the target digital object.

[0082] In some embodiments of the present application, the direct description features are input into the first proxy model, and the thought chain engineering performed by the model is used to infer the potential description features of the target digital object. Thought chain engineering is a technology used for complex reasoning and analysis. Through a multi-step thinking process, the model can infer possible potential description features based on the input direct description features. These potential description features are further abstractions of the implicit needs and features of the target digital object, which helps the system to more comprehensively understand and meet user needs.

[0083] For example, the user's direct descriptive feature is "the latest machine learning algorithm literature". The system inputs this direct descriptive feature into the first agent model, and the model infers potential descriptive features through thought chain engineering analysis, such as "the application of deep learning in medicine" and "comparative study of machine learning algorithm performance". These potential descriptive features represent the implicit needs and additional information that the user may be interested in, allowing the system to provide more comprehensive and accurate data matching results in subsequent steps.

[0084] Sub-step 2033, data fusion is performed on the direct description features and the potential description features to generate demand features.

[0085] In some embodiments of the present application, a demand feature is generated based on the fusion of direct description features and potential description features. The demand feature is a comprehensive description of the target digital object, combining the user's explicit needs and the implicit needs inferred by the system. By integrating direct description features and potential description features, the system is able to generate a comprehensive demand feature, making the subsequent retrieval and matching process more accurate and relevant. This step is intended to improve the system's ability to understand and respond to user needs, thereby providing more effective solutions.

[0086] For example, if the user inputs a requirement description statement of "need the latest literature on machine learning algorithms", the direct description features generated by the system include "machine learning algorithms", "latest literature", etc., and the potential description features include "application of deep learning in medicine", "algorithm performance comparison", etc. Then the system can integrate these direct description features and potential description features to generate a comprehensive requirement feature, such as "latest machine learning algorithms and their applications in medicine", so as to perform more accurate retrieval and matching in subsequent steps. In this way, the system can fully and accurately understand and meet the needs of users.

[0087] Step 204 , searching in the digital tag set according to the demand characteristics, so as to obtain a target digital tag matching the demand characteristics from the digital tag set.

[0088] Each digital tag has a corresponding digital object.

[0089] The method shown in this step has been explained in step 102 and will not be repeated here.

[0090] Optionally, step 204 includes the following sub-steps:

[0091] Sub-step 2041, inputs the demand characteristics into the second agent model, so as to generate a retrieval scheme for the target digital object by obtaining a target retrieval method through selection by the second agent model from a plurality of preset retrieval methods, and searches among a plurality of digital tags according to the retrieval scheme, so as to add the digital tags obtained by the retrieval as candidate digital tags for the target digital object.

[0092] In some embodiments of the present application, the demand characteristics are input into the second proxy model, and the model is used to select from multiple preset retrieval methods to obtain the target retrieval method and generate a retrieval scheme for the target digital object. The second proxy model determines the most suitable retrieval method by analyzing the demand characteristics. The generated retrieval scheme includes specific retrieval steps and methods. The system searches among multiple digital tags according to this scheme and finally determines the candidate digital tags. This step is intended to optimize the retrieval strategy and improve the efficiency and accuracy of the retrieval.

[0093] For example, the user's demand feature is "the latest research on environmental protection". The system inputs this demand feature into the second agent model, and the model selects a target retrieval method based on vector retrieval. The generated retrieval scheme includes the use of a vector retrieval algorithm to first vectorize the demand feature and then match it in the set of digital tags in the database. Based on this retrieval scheme, the system searches through multiple digital tags and eventually determines a set of alternative digital tags, such as "Environmental Protection Research in 2023", "The Latest Ecological Protection Technology", etc. In this way, the system can efficiently determine alternative digital tags that meet user needs.

[0094] Sub-step 2042, inputting the candidate digital labels and the retrieval scheme into the evaluation agent model to obtain retrieval evaluation data for the retrieval scheme and label evaluation data for the candidate digital labels.

[0095] In some embodiments of the present application, the candidate digital tags and the retrieval scheme are input into the evaluation proxy model. The role of the evaluation proxy model is to evaluate the generated retrieval scheme and to evaluate the candidate digital tags. Through this process, the model generates retrieval evaluation data and label evaluation data. The retrieval evaluation data is used to evaluate the effectiveness and efficiency of the current retrieval scheme, and the label evaluation data is used to measure the matching degree between the candidate digital tags and the demand characteristics. Through this step, the system can optimize the retrieval process and improve the accuracy and relevance of the final matching results.

[0096] For example, if the user's demand feature is "the application of artificial intelligence in education", the system generates a set of candidate digital labels, such as "AI educational tools", "machine learning course design", "intelligent teaching platform", etc., and formulates a retrieval scheme based on knowledge graph retrieval. Then the system can input these candidate digital labels and retrieval schemes into the evaluation agent model, and the model will evaluate the retrieval scheme and generate retrieval evaluation data, such as "retrieval coverage", "retrieval speed", etc. At the same time, it will evaluate the candidate digital labels and generate label evaluation data, such as "label relevance score", "label matching degree", etc. Through these evaluation data, the system can judge the effectiveness of the current retrieval scheme and candidate digital labels, and make further optimizations.

[0097] Sub-step 2043, when the tag evaluation data of the candidate digital tag matches the preset tag evaluation condition, the candidate digital tag is determined as the target digital tag.

[0098] In some embodiments of the present application, when the tag evaluation data of the candidate digital tags matches the preset tag evaluation conditions, these candidate digital tags will be determined as target digital tags. The preset tag evaluation conditions are the standards used by the system to measure the matching degree between the tags and the required characteristics. By comparing the tag evaluation data and the preset evaluation conditions, the system can determine which candidate digital tags meet the requirements and determine them as target digital tags to ensure the accuracy and relevance of the retrieval results.

[0099] For example, if the user's demand feature is "smart agricultural technology", the system generates some candidate digital tags, such as "precision agriculture", "smart irrigation system", "agricultural Internet of Things", etc., and evaluates these tags to generate tag evaluation data. If the evaluation data of a candidate digital tag, such as "relevance score" and "matching degree", meets or exceeds the preset tag evaluation conditions (for example, the relevance score reaches more than 80%), the system will identify these tags as target digital tags. In this way, the system can screen out the digital tags that best meet the user's needs and improve the accuracy and effectiveness of the search results.

[0100] Sub-step 2044 , when the search evaluation data of the search scheme does not match the preset search evaluation conditions, initialize the search scheme, adjust the second proxy model according to the search evaluation data, and return to sub-step 2041 .

[0101] In some embodiments of the present application, when the search evaluation data of the search scheme does not match the preset search evaluation conditions, the search scheme is initialized, the second proxy model is adjusted according to the search evaluation data, and the process returns to sub-step 2041. The preset search evaluation conditions are standards for evaluating the effectiveness of the search scheme. If the search evaluation data does not meet these standards, the system will abandon the original search scheme and adjust the second proxy model according to these evaluation data. Through this iterative process, the system can continuously optimize the search model and scheme, thereby gradually improving the efficiency and accuracy of the search.

[0102] For example, the user's demand feature is "intelligent manufacturing technology". The system generated a retrieval scheme based on the BM25 algorithm and searched through the scheme, but the retrieval evaluation data such as the "retrieval accuracy score" was lower than the preset standard (for example, lower than 70%). In sub-step 2044, the system will abandon this retrieval scheme that does not meet the requirements and input the relevant parameters into the evaluation agent model. After model analysis, it will decide how to adjust the second agent model, such as by changing the retrieval algorithm or adjusting the retrieval parameters. Then, the system executes sub-step 2041 again and searches again through the optimized retrieval scheme. In this way, through continuous iteration and optimization, the system can gradually improve the retrieval effect and ultimately meet user needs.

[0103] It should be emphasized that, although in sub-step 2044, the search scheme that did not meet the requirements was initialized, in some specific implementations of the present application, the candidate digital labels retrieved by the previous search scheme will still be retained, thus ensuring that the candidate digital labels are gradually added through the cyclic iteration process. This also ensures that the full performance of the search for the candidate digital labels will not be lost due to the change of the search scheme during the iteration process.

[0104] In some specific embodiments, the adjustment of the second proxy model can be performed through the process of reranking the model (rerankModel). The rerankModel process is a method of reranking the search results using a large language model, which improves the accuracy and efficiency of the search by analyzing and adjusting the relevance of the search results. When the search evaluation data of the search scheme does not meet the preset conditions, the system can optimize the second proxy model through the rerankModel process, thereby improving the subsequent search results.

[0105] Optionally, in some specific embodiments of the present application, the retrieval method may include one or more of vector retrieval, best match (BM) retrieval, knowledge graph retrieval, hybrid retrieval, and web retrieval.

[0106] Step 205: determine the digital object matching the target digital tag as the target digital object.

[0107] The method shown in this step has been explained in step 103 and will not be repeated here.

[0108] Optionally, when there are multiple target digital tags, step 205 includes the following sub-steps:

[0109] Sub-step 2051: determining a digital object corresponding to any one of the multiple target digital tags as a target digital object.

[0110] In some embodiments of the present application, when there are multiple target digital tags, a digital object corresponding to any one of the multiple target digital tags is determined as the target digital object. This step ensures that even if there are multiple possible tags, the digital objects corresponding to these tags can be effectively selected as the final matching target digital objects. This method allows the system to flexibly process multiple matching results and quickly determine the most relevant digital object to meet the needs of users.

[0111] For example, the user's demand feature is "the latest technology of intelligent transportation system". The system generates multiple target digital tags, such as "intelligent traffic management", "autonomous driving technology", "traffic data analysis", etc. At this time, the system can determine the digital object corresponding to any of these target digital tags as the target digital object, such as a research report on "the application of autonomous driving in intelligent transportation" or a technical article on "traffic data analysis". In this way, the system can quickly determine the final target digital object among multiple possible matching results, improving retrieval efficiency and matching accuracy.

[0112] Optionally, when there are multiple target digital objects, the present application further includes the following additional steps:

[0113] Step 206: sort each target digital object according to the number of matches with different target digital tags, and output multiple target digital objects in the sorted order.

[0114] In some embodiments of the present application, when there are multiple target digital objects, each target digital object is sorted according to the number of matches with different target digital tags, and multiple target digital objects are output in the order obtained by sorting. This process is intended to prioritize digital objects by counting how many target digital tags each digital object matches. The basis for sorting is the number of matches between the digital object and the target digital tag. The digital object with more matches has a higher priority, thereby ensuring that the most relevant digital object can be presented to the user first.

[0115] For example, when processing the user's demand for "the latest artificial intelligence applications", the system generates multiple target digital tags, such as "AI medical applications", "AI educational tools", "smart home", etc., and finds multiple digital objects matching these tags. In step 206, the system counts how many target digital tags each digital object matches. For example, an article about "AI applications in medical diagnosis" matches "AI medical applications" and "smart home", while another article about "AI applications in education" only matches "AI educational tools". The system sorts these digital objects according to the number of matches, and prioritizes the output of articles with a larger number of matching tags to ensure that users get the most relevant and comprehensive information.

[0116] Figure 3 This is a complete program implementation process under the method provided in the embodiment of the present application, which shows the whole process from the user inputting a description statement into the system to finally obtaining the displayed search results:

[0117] S1.1 performs preliminary processing on the digital objects to be retrieved and extracts their descriptive features.

[0118] S1.2 The system generates demand features based on the user's description statement, including direct description features and potential description features. Direct description features are the needs clearly expressed by the user, and potential description features are the additional information that the system infers that the user may need.

[0119] S2.1 The user inputs a description of the requirements for the target digital object through natural language, and this input will serve as the basis for the system to understand the user's requirements.

[0120] S2.2 The system analyzes the demand characteristics and selects the most suitable retrieval method, such as vector retrieval, BM retrieval, knowledge graph retrieval, etc.

[0121] S2.3 The system generates a search plan based on the selected search method, and describes the search steps and methods in detail.

[0122] S2.4 The system inputs the candidate digital labels and retrieval schemes into the evaluation agent model and performs evaluation based on the retrieval evaluation data and the label evaluation data.

[0123] S2.5 Determine whether the search plan needs further modification based on the evaluation results.

[0124] S2.5.1 If the search solution has met the requirements, the search results will be fed back.

[0125] S2.5.2 If the search plan has not yet met the requirements, the search plan will be dynamically revised to enter a new round of search process.

[0126] S2.6 The system displays the finalized target digital label to the user.

[0127] In summary, in the embodiments of the present application, by generating demand features, not only based on explicit descriptions, but also based on direct descriptions, potential demands can be inferred, thereby improving the comprehensiveness and accuracy of demand understanding; then, based on the demand features, a search is performed in the digital tag set to obtain a target digital tag that matches the demand features, and the demand features are used to search in the digital tag set, so that the search results are more accurate and relevant. By utilizing the use of digital tags, digital objects can be better characterized, and fast and accurate matching can be achieved, thereby overcoming the inefficiency of traditional manual search and simple keyword matching; finally, the target digital object is determined by matching the target digital tag, which can quickly lock the most relevant data object, improve matching efficiency and accuracy, and make full use of data resources. Therefore, based on the method of the embodiment of the present application, the demand features and digital tags are generated by a large model for centralized retrieval, which reduces the time for users to manually search and automates the complex data matching process; and utilizes the inference and intelligent matching mechanism of potential descriptive data to enable the system to mine and utilize more high-value data resources and avoid missing relevant data. Ultimately, the entire data matching process is automatically executed by the system, and users do not need to intervene frequently, which simplifies operations and improves the accuracy and real-time nature of matching, thereby enhancing user experience and solving the problem of insufficient efficiency and accuracy of data matching in the data space in the prior art.

[0128] refer to Figure 4 , which shows a data space data resource supply and demand matching device 30 based on a large model provided in an embodiment of the present application, comprising:

[0129] The feature generation module 301 is used to determine the requirement features of the target digital object in response to an input operation of a requirement description statement of the target digital object, based on direct description features extracted from the requirement description statement and potential description features obtained by inference operation from the direct description features;

[0130] The tag retrieval module 302 is used to search in the digital tag set according to the demand characteristics to obtain a target digital tag matching the demand characteristics from the digital tag set; each digital tag has a corresponding digital object;

[0131] The digital object module 303 is used to determine the digital object matching the target digital tag as the target digital object.

[0132] Optionally, the feature generation module 301 includes:

[0133] A vectorization submodule, for generating direct description features of a target digital object in response to an input operation of a requirement description statement;

[0134] an inference submodule, for inputting the direct description features into the first proxy model to perform inference operations through the thought chain engineering executed by the first proxy model to obtain potential description features of the target digital object;

[0135] The vector expansion submodule is used to fuse the direct description features and potential description features to generate demand features.

[0136] Optionally, the tag retrieval module 302 includes:

[0137] A retrieval submodule is used to input the demand feature into the second proxy model, so as to generate a retrieval scheme for the target digital object by selecting a target retrieval method from a plurality of preset retrieval methods by the second proxy model, and to search among a plurality of digital tags according to the retrieval scheme, so as to add the digital tag obtained by the retrieval as a candidate digital tag for the target digital object;

[0138] An evaluation submodule, used for inputting candidate digital labels and retrieval schemes into an evaluation proxy model to obtain retrieval evaluation data for the retrieval schemes and label evaluation data for the candidate digital labels;

[0139] A selection submodule, for determining the candidate digital label as the target digital label when the label evaluation data of the candidate digital label matches the preset label evaluation condition;

[0140] The iterative submodule is used to initialize the retrieval scheme when the retrieval evaluation data of the retrieval scheme does not match the preset retrieval evaluation conditions, adjust the second proxy model according to the retrieval evaluation data, and return to the step of inputting the demand characteristics into the second proxy model to obtain the target retrieval method through the selection of the second proxy model among the preset multiple retrieval methods, generate a retrieval scheme for the target digital object, and search among multiple digital tags according to the retrieval scheme to add the digital tags obtained by the retrieval as the candidate digital tags of the target digital object.

[0141] Optionally, when there are multiple target digital tags, the digital object module 303 includes:

[0142] A digital object submodule, used to determine a digital object corresponding to any one of the multiple target digital tags as a target digital object;

[0143] The data space data resource supply and demand matching device 30 based on the large model also includes:

[0144] The sorting module is used to sort each target digital object according to the number of matches with different target digital labels, and output multiple target digital objects in the sorted order.

[0145] Optionally, the data space data resource supply and demand matching device 30 based on the large model further includes:

[0146] A description data module, used for determining description data of a digital object;

[0147] The digital label module is used to generate a digital label of the digital object according to the description data of the digital object, and store the digital object in the database according to the generated digital label.

[0148] Optionally, the description data module includes:

[0149] A text description submodule, for performing text recognition on the digital object when the digital object is text data, so as to use the knowledge graph of the digital object obtained through text recognition as description data;

[0150] The non-structure description submodule is used to perform content recognition on the digital object when the digital object is non-structured data, so as to use the description text of the digital object obtained by content recognition as description data;

[0151] The multimodal description submodule is used to perform content recognition on sub-data in different modes in the digital object respectively when the digital object is multimodal data, so as to obtain multiple description sub-texts of the digital object through content recognition, and generate description data based on the obtained multiple description sub-texts.

[0152] Optionally, the multimodal description submodule includes:

[0153] A question unit is used to input multiple description subtexts into the question guessing model respectively to obtain a question subtext for the digital object; the question subtext is a description statement of the requirements for the digital object obtained by guessing based on the description subtext;

[0154] The multimodal description unit is used to use the description subtext and / or the question subtext as description data.

[0155] Optional digital label module includes:

[0156] A matching submodule, used for taking each description data of the digital object as a digital label of the digital object;

[0157] The index submodule is used to store all digital objects in the database and establish a corresponding relationship between digital tags and digital objects in the database according to all digital tags.

[0158] In summary, in the embodiments of the present application, by generating demand features, not only based on explicit descriptions, but also based on direct descriptions, potential demands can be inferred, thereby improving the comprehensiveness and accuracy of demand understanding; then, based on the demand features, a search is performed in the digital tag set to obtain a target digital tag that matches the demand features, and the demand features are used to search in the digital tag set, so that the search results are more accurate and relevant. By utilizing the use of digital tags, digital objects can be better characterized, and fast and accurate matching can be achieved, thereby overcoming the inefficiency of traditional manual search and simple keyword matching; finally, the target digital object is determined by matching the target digital tag, which can quickly lock the most relevant data object, improve matching efficiency and accuracy, and make full use of data resources. Therefore, based on the method of the embodiment of the present application, the demand features and digital tags are generated by a large model for centralized retrieval, which reduces the time for users to manually search and automates the complex data matching process; and utilizes the inference and intelligent matching mechanism of potential descriptive data to enable the system to mine and utilize more high-value data resources and avoid missing relevant data. Ultimately, the entire data matching process is automatically executed by the system, and users do not need to intervene frequently, which simplifies operations and improves the accuracy and real-time nature of matching, thereby enhancing user experience and solving the problem of insufficient efficiency and accuracy of data matching in the data space in the prior art.

[0159] Reference Figure 5 , the electronic device 500 may include one or more of the following components: a processing component 502 , a memory 504 , a power component 506 , a multimedia component 508 , an audio component 510 , an input / output (I / O) interface 512 , a sensor component 514 , and a communication component 516 .

[0160] The processing component 502 generally controls the overall operation of the electronic device 500, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 502 may include one or more processors 520 to execute instructions to complete all or part of the steps of the above-mentioned method. In addition, the processing component 502 may include one or more modules to facilitate the interaction between the processing component 502 and other components. For example, the processing component 502 may include a multimedia module to facilitate the interaction between the multimedia component 508 and the processing component 502.

[0161] The memory 504 is used to store various types of data to support the operation of the electronic device 500. Examples of such data include instructions for any application or method operating on the electronic device 500, contact data, phone book data, messages, pictures, multimedia, etc. The memory 504 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0162] The power supply component 506 provides power to the various components of the electronic device 500. The power supply component 506 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device 500.

[0163] The multimedia component 508 includes an interface that provides an output interface between the electronic device 500 and the user. In some embodiments, the interface may include a liquid crystal display (LCD) and a touch panel (TP). If the interface includes a touch panel, the interface may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 508 includes a front camera and / or a rear camera. When the electronic device 500 is in an operating mode, such as a shooting mode or a multimedia mode, the front camera and / or the rear camera may receive external multimedia data. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

[0164] The audio component 510 is used to output and / or input audio signals. For example, the audio component 510 includes a microphone (MIC), and when the electronic device 500 is in an operation mode, such as a call mode, a recording mode, and a voice recognition mode, the microphone is used to receive an external audio signal. The received audio signal can be further stored in the memory 504 or sent via the communication component 516. In some embodiments, the audio component 510 also includes a speaker for outputting audio signals.

[0165] The input / output I / O interface 512 provides an interface between the processing component 502 and the peripheral interface module, which may be a keyboard, a click wheel, buttons, etc. These buttons may include but are not limited to: a home button, a volume button, a start button, and a lock button.

[0166] The sensor assembly 514 includes one or more sensors for providing various aspects of status assessment for the electronic device 500. For example, the sensor assembly 514 can detect the open / closed state of the electronic device 500, the relative positioning of components, such as the display and keypad of the electronic device 500, and the sensor assembly 514 can also detect the position change of the electronic device 500 or a component of the electronic device 500, the presence or absence of user contact with the electronic device 500, the orientation or acceleration / deceleration of the electronic device 500, and the temperature change of the electronic device 500. The sensor assembly 514 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 514 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 514 may also include an accelerometer, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0167] The communication component 516 is used to facilitate wired or wireless communication between the electronic device 500 and other devices. The electronic device 500 can access a wireless network based on a communication standard, such as WiFi, a carrier network (such as 2G, 3G, 4G or 5G), or a combination thereof. In an exemplary embodiment, the communication component 516 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 516 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0168] In an exemplary embodiment, the electronic device 500 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to implement the methods provided in the embodiments of the present application.

[0169] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 504 including instructions, and the instructions can be executed by a processor 520 of an electronic device 500 to perform the above method. For example, the non-transitory storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0170] Figure 6FIG. 6 is a block diagram of an electronic device 600 according to another embodiment of the present invention. For example, the electronic device 600 may be provided as a server.

[0171] Reference Figure 6 , the electronic device 600 includes a processing component 622, which further includes one or more processors, and a memory resource represented by a memory 632, for storing instructions that can be executed by the processing component 622, such as an application. The application stored in the memory 632 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 622 is configured to execute instructions to perform the method provided in the embodiment of the present application.

[0172] The electronic device 600 may also include a power supply component 626 configured to perform power management of the electronic device 600, a wired or wireless network interface 650 configured to connect the electronic device 600 to a network, and an input / output (I / O) interface 658. The electronic device 600 may operate based on an operating system stored in the memory 632, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™ or the like.

[0173] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the application disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0174] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A data space data resource supply and demand matching method based on a large model, applied to paper retrieval, characterized in that: include: In response to an input operation of a requirement description statement of a target digital object, determining the requirement features of the target digital object according to direct description features extracted from the requirement description statement and potential description features obtained by inference operation from the direct description features; The target digital object is a paper corresponding to the requirement description statement; According to the requirement characteristics, searching in the digital tag set to obtain a target digital tag matching the requirement characteristics from the digital tag set; each of the digital tags has a corresponding digital object; Determine a digital object matching the target digital tag as the target digital object; The searching in the digital tag set according to the requirement feature to obtain a target digital tag matching the requirement feature from the digital tag set includes: Inputting the demand feature into the second proxy model in the macro model, generating a search scheme for the target digital object by obtaining a target search method through selection by the second proxy model from among a plurality of preset search methods, and searching among the plurality of digital tags according to the search scheme, so as to add the digital tag obtained by the search as a candidate digital tag for the target digital object; Inputting the candidate digital labels and the retrieval scheme into an evaluation agent model to obtain retrieval evaluation data for the retrieval scheme and label evaluation data for the candidate digital labels; In the case where the label evaluation data of the candidate digital label matches the preset label evaluation condition, determining the candidate digital label as the target digital label; In the case that the retrieval evaluation data of the retrieval scheme does not match the preset retrieval evaluation conditions, the retrieval scheme is initialized, and the second agent model is adjusted according to the retrieval evaluation data, and the step of inputting the demand characteristics into the second agent model is returned to generate a retrieval scheme for the target digital object by obtaining a target retrieval method through the selection of the second agent model among the preset multiple retrieval methods, and according to the retrieval scheme, a search is performed among the multiple digital tags to add the digital tags obtained by the retrieval as candidate digital tags for the target digital object.

2. The data space data resource supply and demand matching method based on a large model as claimed in claim 1, characterized in that: In response to the input operation of the requirement description statement of the target digital object, the requirement features of the target digital object are determined according to the direct description features extracted from the requirement description statement and the potential description features obtained by inference operation from the direct description features, including: In response to an input operation of the requirement description statement, generating a direct description feature of the target digital object; Inputting the direct description features into the first proxy model in the large model, so as to perform reasoning operation through the thought chain engineering executed by the first proxy model, so as to obtain the potential description features of the target digital object; The direct description feature and the potential description feature are data-fused to generate the demand feature.

3. The data space data resource supply and demand matching method based on a large model as claimed in claim 1, characterized in that: In the case that there are multiple target digital tags, determining the digital object matching the target digital tag as the target digital object includes: Determine a digital object corresponding to any one of the plurality of target digital tags as the target digital object; In the case where there are multiple target digital objects, the data space data resource supply and demand matching method based on the big model further includes: Each of the target digital objects is sorted according to the number of matches with different target digital tags, and a plurality of the target digital objects are output in the sorted order.

4. The method for matching supply and demand of data space data resources based on a large model as claimed in claim 1, characterized in that: The data space data resource supply and demand matching method based on the big model also includes: Determining description data of the digital object; A digital tag of the digital object is generated according to the description data of the digital object, and the digital object is stored in a database according to the generated digital tag.

5. The data space data resource supply and demand matching method based on a large model as claimed in claim 4, characterized in that: Determining the description data of the digital object includes: In the case where the digital object is text data, performing text recognition on the digital object, so as to use the knowledge graph of the digital object obtained by the text recognition as the description data; In the case where the digital object is unstructured data, performing content recognition on the digital object, so as to use a description text of the digital object obtained by the content recognition as the description data; In the case where the digital object is multimodal data, content recognition is performed on sub-data in different modalities of the digital object respectively, so as to obtain multiple description sub-texts of the digital object through the content recognition, and the description data is generated according to the obtained multiple description sub-texts.

6. The method for matching supply and demand of data space data resources based on a large model as claimed in claim 4, characterized in that: Generating a digital tag of the digital object according to the description data of the digital object, and storing the digital object in a database according to the generated digital tag, comprises: Using each description data of the digital object as a digital label of the digital object; All of the digital objects are stored in a database, and a corresponding relationship between the digital tags and the digital objects is established in the database according to all of the digital tags.

7. A data space data resource supply and demand matching device based on a large model, applied to paper retrieval, characterized in that: include: A feature generation module, configured to determine the requirement features of the target digital object in response to an input operation of a requirement description statement of the target digital object, based on direct description features extracted from the requirement description statement and potential description features obtained through inference operations from the direct description features; The target digital object is a paper corresponding to the requirement description statement; A tag retrieval module, used to search in a digital tag set according to the demand characteristics, so as to obtain a target digital tag matching the demand characteristics from the digital tag set; each of the digital tags has a corresponding digital object; A digital object module, configured to determine a digital object matching the target digital tag as the target digital object; The tag retrieval module includes: A retrieval submodule, configured to input the demand feature into the second proxy model in the large model, to generate a retrieval scheme for the target digital object by selecting a target retrieval method from a plurality of preset retrieval methods by the second proxy model, and to search among the plurality of digital tags according to the retrieval scheme, so as to add the digital tag obtained by the retrieval as a candidate digital tag for the target digital object; An evaluation submodule, used for inputting the candidate digital labels and the retrieval scheme into an evaluation agent model to obtain retrieval evaluation data for the retrieval scheme and label evaluation data for the candidate digital labels; A selection submodule, configured to determine the candidate digital label as a target digital label when the label evaluation data of the candidate digital label matches a preset label evaluation condition; An iterative submodule is used to initialize the retrieval scheme when the retrieval evaluation data of the retrieval scheme does not match the preset retrieval evaluation conditions, and adjust the second proxy model according to the retrieval evaluation data, and return to the step of inputting the demand characteristics into the second proxy model to generate a retrieval scheme for the target digital object by obtaining a target retrieval method through the selection of the second proxy model among multiple preset retrieval methods, and according to the retrieval scheme, search among the multiple digital tags to add the digital tags obtained by the retrieval as candidate digital tags for the target digital object.

8. An electronic device, characterized in that: include: A processor, a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the data space data resource supply and demand matching method based on a large model as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the data space data resource supply and demand matching method based on a large model as described in any one of claims 1 to 6.

Images