Data asset tag generation method, apparatus, device, medium, and program product

Abstract

This application provides a data asset tag generation method, applicable to the field of artificial intelligence technology, specifically involving the application of large-scale models in the fintech field. The data asset tag generation method includes: obtaining user data tag requirements for managing data assets; inputting the data tag requirements into a pre-built multimodal tag automatic generation model, and performing the following operations: determining the tag type corresponding to the data tag requirements; extracting knowledge fragments corresponding to the tag type from a pre-built knowledge base, and generating contextual prompts based on the knowledge fragments, wherein the knowledge base is pre-divided into multiple types of topic domains; generating candidate tags based on the knowledge fragments and contextual prompts; and tagging data assets that meet preset conditions based on the candidate tags. This application also provides a data asset tag generation apparatus, device, medium, and program product.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, specifically to the application of large models in the field of financial technology, and more specifically to a method, apparatus, device, medium and program product for generating data asset tags. Background Technology

[0002] As financial institutions' data assets grow increasingly large, the phenomenon of data silos becomes more and more obvious. The existing data asset architecture from a technological perspective can no longer meet business needs. Relevant technical personnel face the challenge of efficiently "finding data," "understanding data," and "using data." Therefore, it is necessary to establish a connection between business and data assets, for example, by using data tags to describe data assets in a more refined way.

[0003] However, under the current technological architecture, existing technologies heavily rely on manual methods to generate and maintain data tags for data assets. This approach is not only inefficient when dealing with massive amounts of highly dynamic data, but also significantly increases the computational overhead within computer systems. Furthermore, the lack of a unified and automated metadata retrieval mechanism means that systems must perform a large amount of redundant calculations and data scanning when searching, understanding, and utilizing data, resulting in a waste of system computing resources. Summary of the Invention

[0004] In view of the above problems, this application provides a method, apparatus, device, medium and program product for generating data asset tags that can reduce system computing overhead and optimize system computing resources.

[0005] According to a first aspect of this application, a data asset tag generation method is provided, comprising: obtaining data tag requirements from users for managing data assets; inputting the data tag requirements into a pre-built multimodal tag automatic generation model, and performing the following operations: determining the tag type corresponding to the data tag requirements; extracting knowledge fragments corresponding to the tag type from a pre-built knowledge base, and generating contextual prompts based on the knowledge fragments, wherein the knowledge base is pre-divided into multiple types of topic domains, and each type of topic domain includes multiple knowledge fragments corresponding to each tag type; generating candidate tags based on the knowledge fragments and contextual prompts; and tagging data assets that meet preset conditions based on the candidate tags.

[0006] According to embodiments of this application, the multimodal tag automatic generation model is built upon a large model. The structure of the multimodal tag automatic generation model includes: a tag type layer, configured to determine the tag type corresponding to the data tag requirements; a knowledge fragment extraction layer, configured to extract knowledge fragments corresponding to the tag type from a pre-built knowledge base and generate contextual prompts based on the knowledge fragments; a tag generation layer, configured to generate candidate tags based on the knowledge fragments and contextual prompts; and a data asset labeling layer, configured to label data assets that meet preset conditions based on the candidate tags.

[0007] According to an embodiment of this application, pre-constructing a knowledge base includes: acquiring raw data information; preprocessing the raw data information and dividing the preprocessed raw data information into multiple topic domain knowledge bases, wherein the knowledge bases are used to provide background knowledge for generating candidate tags; and generating a knowledge graph with multiple topic domains and a vector index corresponding to the knowledge graph using a knowledge fusion model based on the multiple topic domain knowledge bases.

[0008] According to embodiments of this application, generating a knowledge graph with multiple topic domains using a knowledge fusion model based on a knowledge base of multiple topics includes: extracting graph relationship features from graph data using a graph neural network based on graph data in the knowledge base; extracting key information features from complex text data using a first text classification model based on complex text data in the knowledge base; extracting word frequency features from simple text data using a second text classification model based on simple text data in the knowledge base; extracting resource attributes and relationship features of each metadata in the structured data using a standard data model based on structured data in the knowledge base; and fusing graph relationship features, key information features, word frequency features, resource attributes, and relationship features to generate a knowledge graph corresponding to each topic domain.

[0009] According to an embodiment of this application, extracting knowledge fragments corresponding to a tag type from a pre-built knowledge base based on the tag type includes: determining the candidate range of the knowledge base based on the tag type, and extracting subgraphs in the knowledge graph corresponding to the tag type within the candidate range; determining candidate topic domains of the subgraphs based on vector indices; calculating the similarity between the tag type and each candidate topic domain within the subgraph, and extracting candidate topic domains with similarity values ​​higher than a first similarity threshold to obtain knowledge fragments.

[0010] According to an embodiment of this application, calculating the similarity between the label type and each candidate topic domain within the subgraph, and extracting candidate topic domains with similarity values ​​higher than a first similarity threshold to obtain knowledge fragments includes: determining the value vector corresponding to each candidate topic domain based on the importance of the candidate topic domain and time update parameters, wherein the value vector represents the semantic knowledge features of each candidate topic domain; calculating the first cosine similarity between the features corresponding to the label type and the semantic knowledge features corresponding to each candidate topic domain; and extracting the semantic knowledge features corresponding to the candidate topic domains with similarity values ​​higher than the first similarity threshold to obtain knowledge fragments.

[0011] According to an embodiment of this application, tagging data assets that meet preset conditions based on candidate tags includes: configuring a tagging mechanism, wherein the tagging mechanism is configured to automatically tag data assets that meet preset conditions with corresponding tags; and tagging data assets that meet preset conditions based on candidate tags using the tagging mechanism.

[0012] According to an embodiment of this application, the method further includes: performing a second cosine similarity detection on data tags of the same type; triggering a tag addition suggestion in response to the second cosine similarity being lower than a second similarity threshold; generating new data tags using a multimodal tag automatic generation model based on the tag addition suggestion; and tagging data assets that meet preset conditions based on the new data tags.

[0013] According to embodiments of this application, the method further includes: calculating the activity level of data tags for each business domain; determining data tags with activity levels higher than an activity threshold as high-activity data tags in response to the activity level being higher than an activity threshold; and recommending data assets corresponding to the high-activity data tags to users in the corresponding business domain based on the high-activity data tags.

[0014] A second aspect of this application provides a data asset tag generation apparatus, comprising: an acquisition module for acquiring data tag requirements from users for managing data assets; inputting the data tag requirements into a pre-built multimodal tag automatic generation model, and executing the following modules: a determination module for determining the tag type corresponding to the data tag requirements; a first generation module for extracting knowledge fragments corresponding to the tag type from a pre-built knowledge base, and generating contextual prompts based on the knowledge fragments, wherein the knowledge base is pre-divided into multiple types of topic domains, and each type of topic domain includes multiple knowledge fragments corresponding to each tag type; a second generation module for generating candidate tags based on the knowledge fragments and contextual prompts; and a tagging module for tagging data assets that meet preset conditions based on the candidate tags.

[0015] A third aspect of this application provides an electronic device comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method described above.

[0016] A fourth aspect of this application also provides a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implement the steps of the above-described method.

[0017] The fifth aspect of this application also provides a computer program product, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described method. Attached Figure Description

[0018] The above-mentioned contents, other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0019] Figure 1 This illustration schematically depicts an application scenario of the data asset tag generation method according to an embodiment of this application.

[0020] Figure 2 A flowchart illustrating a data asset tag generation method according to an embodiment of this application is shown schematically;

[0021] Figure 3 A flowchart illustrating a method for pre-building a knowledge base according to an embodiment of this application is shown schematically.

[0022] Figure 4 This illustration schematically shows a flowchart of a method for generating a knowledge graph with multiple subject domains using a knowledge fusion model according to an embodiment of this application;

[0023] Figure 5 This illustration schematically shows a flowchart of a method for extracting knowledge fragments from a pre-built knowledge base based on tag type according to an embodiment of this application;

[0024] Figure 6 This illustration schematically shows a flowchart of a method for extracting knowledge fragments based on the similarity between tag type and candidate topic domains according to an embodiment of this application;

[0025] Figure 7 This illustration schematically shows a flowchart of a method for automatic labeling based on candidate labels according to an embodiment of this application;

[0026] Figure 8 This schematically illustrates a structural block diagram of a data asset tag generation apparatus according to an embodiment of this application; and

[0027] Figure 9A block diagram schematically illustrates an electronic device suitable for implementing a data asset tag generation method according to an embodiment of this application. Detailed Implementation

[0028] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0029] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0030] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0031] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0032] It should be noted that the data asset tag generation method and apparatus of this application can be used in the field of fintech for data asset tag generation, and can also be used in any field other than fintech for data asset tag generation. The application field of the data asset tag generation method and apparatus of this application is not limited.

[0033] In the technical solution of this application, the user information (including but not limited to user personal information, user image information, user device information, such as location information) and data (including but not limited to data used for analysis, stored data, and displayed data) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of related data all comply with relevant laws, regulations, and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse.

[0034] In scenarios involving automated decision-making using personal information, the methods, devices, and systems provided in this application all offer users corresponding entry points for choosing to agree to or reject the automated decision-making results. If the user chooses to reject, the process proceeds to the expert decision-making stage. Here, "automated decision-making" refers to the activity of automatically analyzing and evaluating an individual's behavioral habits, interests, or economic, health, and credit status through computer programs, and then making a decision. Here, "expert decision-making" refers to the activity of making decisions by personnel who specialize in a particular field, possess specialized experience, knowledge, and skills, and have reached a certain level of professional expertise.

[0035] Embodiments of this application provide a data asset tag generation method, including: obtaining data tag requirements from users for managing data assets; inputting the data tag requirements into a pre-built multimodal tag automatic generation model, and performing the following operations: determining the tag type corresponding to the data tag requirements; extracting knowledge fragments corresponding to the tag type from a pre-built knowledge base, and generating contextual prompts based on the knowledge fragments, wherein the knowledge base is pre-divided into multiple types of topic domains, and each type of topic domain includes multiple knowledge fragments corresponding to each tag type; generating candidate tags based on the knowledge fragments and contextual prompts; and tagging data assets that meet preset conditions based on the candidate tags.

[0036] Figure 1 The diagram illustrates an application scenario of the data asset tag generation method according to an embodiment of this application.

[0037] like Figure 1 As shown, application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, and a third terminal device 103. Network 104 serves as a medium for providing communication links between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. Network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0038] Users can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 via the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).

[0039] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0040] Server 105 can be a server that provides various services, such as a backend management server that supports websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (this is just an example). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0041] It should be noted that the data asset tag generation method provided in this application embodiment can generally be executed by server 105. Correspondingly, the data asset tag generation device provided in this application embodiment can generally be located in server 105. The data asset tag generation method provided in this application embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105. Correspondingly, the data asset tag generation device provided in this application embodiment can also be located in a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105.

[0042] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0043] The following will be based on Figure 1 The described scene, through Figures 2-7 The data asset tag generation method according to the embodiments of this application will be described in detail.

[0044] Figure 2 A flowchart illustrating a data asset tag generation method according to an embodiment of this application is shown.

[0045] like Figure 2 As shown, the data asset tag generation method of this embodiment includes operations S210 to S250, and the data asset tag generation method can be executed by a server.

[0046] In operation S210, the user's data tag requirements for managing data assets are obtained.

[0047] Input the data label requirements into the pre-built multimodal label to automatically generate the model, and perform the following operations S220 to S250.

[0048] In operation S220, the label type corresponding to the data label requirements is determined based on the data label requirements.

[0049] In operation S230, knowledge fragments corresponding to the tag type are extracted from the pre-built knowledge base according to the tag type, and contextual prompts are generated based on the knowledge fragments. The knowledge base is pre-divided into multiple types of topic domains, and each type of topic domain includes multiple knowledge fragments corresponding to each tag type.

[0050] In operation S240, candidate tags are generated based on knowledge fragments and contextual prompts.

[0051] In operation S250, data assets that meet preset conditions are tagged based on candidate tags.

[0052] In the embodiments of this application, the data tagging requirements are first obtained. Data tagging requirements are the specific requirements of users for managing data assets (such as classifying, describing or evaluating them), and are usually related to business scenarios, technical goals or management needs.

[0053] For example, the data labeling requirement could be "evaluating data asset quality labels" or "designing personal business domain scenario labels."

[0054] "Evaluate data asset quality labels" requires the generation of labels related to data quality (such as "data integrity", "data accuracy", etc.).

[0055] "Design scenario labels for personal banking services" requires generating scenario-based labels for personal banking services (such as "personal financial management" and "customer service").

[0056] Then, the data label requirements are input into a pre-built multimodal label automatic generation model to automatically classify, generate, and label the labels.

[0057] After the data tagging requirements are input into the multimodal tag generation model, the model can first determine the tag type corresponding to the input data tagging requirements. For example, after inputting data tagging requirements, the first layer can quickly determine whether the data tagging requirements belong to the business, technology, management, or operations category.

[0058] Then, based on the type of data tags, the scope of the knowledge base is automatically limited (e.g., based on the type of data tags, the topic domain of the knowledge base is quickly located). High-value knowledge fragments corresponding to the tag types are then extracted from the topic domain, and contextual prompts are generated based on the extracted knowledge fragments.

[0059] This knowledge base is pre-built based on the original data information and is divided into multiple subject domains. Typically, the number of subject domains in the knowledge base corresponds to the number of data tag types. Each subject domain contains multiple knowledge fragments corresponding to each tag type, which are used to provide data sources for subsequent tag generation.

[0060] This multimodal label generation model can further generate candidate labels (such as "data integrity", "data accuracy", etc.) based on high-value knowledge fragments and contextual prompts.

[0061] Finally, this multimodal labeling automatic generation model can automatically label data assets that meet preset conditions according to preset labeling rules. These labeling rules can be understood as "what conditions data assets should meet to be labeled with this type of candidate label".

[0062] According to embodiments of this application, by pre-constructing a professional database, when given data tagging requirements (constraints), these requirements are input into a multimodal tag generation model. The model can quickly locate the corresponding topic domain in the knowledge base based on the type of data tagging requirements and extract the knowledge fragments needed to generate tags, thereby forming candidate tags for automatic tagging of data assets. Since the entire process of tag classification, tag generation, and tag printing is automatically completed by the multimodal tag generation model, users only need to input data tagging requirements to intelligently and quickly locate the relevant topic domain in the knowledge base based on the type of data tagging requirements, and efficiently extract the key knowledge fragments needed to generate data tags, significantly narrowing the information retrieval scope and reducing computational complexity. Simultaneously, because it can accurately extract knowledge fragments from the knowledge base, redundant calculations and full-database scanning are reduced, thus greatly optimizing the utilization efficiency of system memory and other computing resources.

[0063] In the embodiments of this application, the multimodal tag automatic generation model is built based on a large model. The structure of the multimodal tag automatic generation model includes: a tag type layer, a knowledge fragment extraction layer, a tag generation layer, and a data asset annotation layer.

[0064] The tag type layer is configured to determine the tag type corresponding to the data tag requirements based on the data tag requirements.

[0065] The knowledge fragment extraction layer is configured to extract knowledge fragments corresponding to the tag type from a pre-built knowledge base, and generate contextual prompts based on the knowledge fragments.

[0066] The tag generation layer is configured to generate candidate tags based on knowledge fragments and contextual prompts.

[0067] The data asset labeling layer is configured to label data assets that meet preset conditions based on candidate labels.

[0068] In this embodiment of the application, a four-layer multimodal label automatic generation model is constructed based on a large model.

[0069] Large models are artificial intelligence models capable of understanding and generating natural language. They simulate human intelligent activities and reasoning processes by learning from massive amounts of text data. Large models possess powerful semantic understanding capabilities; they can not only identify keywords but also understand the deeper meaning of text, context, and user intent. This enables them to handle complex natural language input and perform accurate classification, summarization, question answering, and other tasks. Therefore, in this embodiment, large models are used to construct a multimodal label automatic generation model. Its four-layer structure is sequentially connected and works in concert to perform fully automated processes for the input data labeling requirements.

[0070] According to embodiments of this application, the large model has powerful semantic understanding capabilities. By constructing a multimodal tag automatic generation model based on the large model, it is possible to quickly and accurately locate user needs and relevant information in the knowledge base, avoiding redundant scanning and complex reasoning of the entire knowledge base, and significantly reducing computational complexity.

[0071] Figure 3 A flowchart illustrating a method for pre-building a knowledge base according to an embodiment of this application is shown.

[0072] like Figure 3 As shown, the method for pre-building a knowledge base in this embodiment includes operations S310 to S330.

[0073] Operate S310 to obtain raw data information.

[0074] In operation S320, the raw data information is preprocessed and the preprocessed raw data information is divided into multiple subject domain knowledge bases. The knowledge bases are used to provide background knowledge for generating candidate tags.

[0075] When operating S330, a knowledge graph with multiple subject domains is generated using a knowledge fusion model based on a knowledge base of multiple subject domains, along with corresponding vector indices.

[0076] In this embodiment of the application, a large amount of raw data information is integrated and divided into knowledge bases of multiple subject domains, and knowledge graphs and vector indexes are constructed based on these to provide background support for subsequent tag generation.

[0077] For example, raw data can be collected from various data sources (such as databases, documents, application programming interfaces (APIs), web pages, etc.) and preprocessed, such as data cleaning, to handle missing values, duplicate data, inconsistent formats, etc., in order to ensure the accuracy of subsequent analysis.

[0078] In this embodiment of the application, four types of topic knowledge bases can be constructed based on the original data information, specifically including:

[0079] Business knowledge base: can be divided into subcategories such as business knowledge of various departments and fields, regulatory requirements, etc., including documents such as industry business terminology tables and product manuals.

[0080] Hot Topics Database: Can be categorized by source, including regulatory clauses, industry news, and other documents.

[0081] Project Case Library: It can be divided into subcategories according to the project's competent authority and requirement type, including historical project documents, business process diagrams, and other types of files.

[0082] Asset Information Repository: It can be divided into subcategories according to asset type, including data table lineage charts, asset metadata information and other types of files.

[0083] Then, using a knowledge fusion model, a knowledge graph and vector index are generated for each topic knowledge base.

[0084] A knowledge graph is a model that uses a graph structure to represent knowledge. Nodes represent entities (such as people, things, concepts, etc.), and edges represent relationships between entities (such as "belonging to", "cooperating", etc.).

[0085] Vector indexing converts data such as text and images into high-dimensional vector representations and stores them in an index structure to support efficient similarity searches later.

[0086] According to embodiments of this application, the topic domains and knowledge fragments in the knowledge base provide structured background knowledge for tag generation. Combined with the reasoning capabilities of the knowledge graph and the semantic search capabilities of the vector index, the generated candidate tags will better meet the semantic requirements.

[0087] Figure 4 The flowchart illustrating a method for generating a knowledge graph with multiple subject domains using a knowledge fusion model according to an embodiment of this application is shown.

[0088] like Figure 4 As shown, the method for generating a knowledge graph with multiple subject domains using a knowledge fusion model in this embodiment includes operations S410 to S450.

[0089] When operating S410, based on the graph data in the knowledge base, a graph neural network is used to extract the graph relationship features in the graph data.

[0090] When operating S420, based on complex text data in the knowledge base, the first text classification model is used to extract key information features from the complex text data.

[0091] When operating S430, based on simple text data in the knowledge base, the second text classification model is used to extract word frequency features from the simple text data.

[0092] When operating S440, based on the structured data in the knowledge base, the resource attributes and relationship characteristics of each metadata in the structured data are extracted using a standard data model.

[0093] By operating the S450, knowledge graphs are generated by integrating graph relationship features, key information features, word frequency features, resource attributes and relationship features, and generating knowledge graphs corresponding to each subject domain.

[0094] In this embodiment, a knowledge fusion model that integrates graph neural networks, a first text classification model, a second text classification model, and a standard data model is used to generate a knowledge graph.

[0095] This knowledge fusion model integrates graph neural networks, a first text classification model, a second text classification model, and a standard data model. It classifies various materials in the knowledge base, extracts features using this knowledge fusion model, and constructs a knowledge graph. Specifically:

[0096] For graph-based data in the knowledge base (such as the lineage graph of the data table in the asset information database), graph neural networks are used to extract the graph relationship features in the graph data.

[0097] For complex text data in the knowledge base (such as regulatory clauses in the hot information database), the first text classification model is used to extract key information features from the complex text data and capture the regulatory focus in the clauses (such as "cross-border transaction reporting requirements").

[0098] For simple text data in the knowledge base (such as the glossary in the business knowledge base), the word frequency features in the simple text data are extracted using the second text classification model.

[0099] For structured data (such as data asset metadata) in the knowledge base, the resource attributes and relationship characteristics of each metadata in the structured data are extracted using standard data models.

[0100] The features extracted from each model are then fused and spliced ​​together to generate a knowledge graph.

[0101] According to the embodiments of this application, since the knowledge fusion model integrates graph neural networks, a first text classification model, a second text classification model, and a standard data model, it can capture multi-dimensional data features in the knowledge base in all aspects, improve the comprehensiveness of knowledge representation, and enhance the deep fusion of semantic and structural relationships, thereby improving the reasoning ability of the knowledge graph. This is beneficial for generating more accurate labels in the future.

[0102] Figure 5 The flowchart illustrates a method for extracting knowledge fragments from a pre-built knowledge base based on tag type, according to an embodiment of this application.

[0103] like Figure 5 As shown, the method for extracting knowledge fragments from a pre-built knowledge base based on tag type in this embodiment includes operations S510 to S530.

[0104] In the S510 operation, the candidate range of the knowledge base is determined according to the tag type, and within the candidate range, the subgraph corresponding to the tag type in the knowledge graph is extracted.

[0105] In operation S520, candidate topic domains for the subgraph are determined based on the vector index.

[0106] In operation S530, the similarity between the label type and each candidate topic domain within the subgraph is calculated, and candidate topic domains with similarity values ​​higher than the first similarity threshold are extracted to obtain knowledge fragments.

[0107] In this embodiment of the application, the knowledge fragments required for generating tags are obtained by calculating the similarity between the tag type features and the candidate topic domain features of the knowledge base.

[0108] First, the scope of the knowledge base is limited according to the type of data tags. For example, for technical data tags, only the asset information base, business knowledge base, and hot topic information base are extracted, and then the corresponding sub-graphs are extracted.

[0109] Then, candidate topic domains for the subgraph are selected through knowledge vector retrieval, also known as vector indexing.

[0110] Finally, the similarity between the label type and each candidate topic domain within the subgraph is calculated, and a first similarity threshold is set. Knowledge fragments are extracted from candidate topic domains that are higher than the first similarity threshold.

[0111] According to the embodiments of this application, knowledge fragments are filtered by calculating the similarity between tag type features and candidate topic domain features, which can significantly improve the relevance and accuracy of knowledge fragments, optimize retrieval efficiency, enhance contextual relevance, and ensure that the extracted knowledge fragments provide high-quality background support for tag generation, making it suitable for complex and diverse business scenarios.

[0112] Figure 6 The flowchart illustrating a method for extracting knowledge fragments based on the similarity between label type and candidate topic domains according to an embodiment of this application is shown.

[0113] like Figure 6 As shown, the method for extracting knowledge fragments based on the similarity between label type and candidate topic domain in this embodiment includes operations S610 to S630.

[0114] In operation S610, the value vector corresponding to each candidate topic domain is determined based on the importance of the candidate topic domain and the time update parameters. The value vector represents the semantic knowledge features of each candidate topic domain.

[0115] In operation S620, the first cosine similarity between the features corresponding to the label type and the semantic knowledge features corresponding to each candidate topic domain is calculated.

[0116] In operation S630, in response to the first cosine similarity being higher than the first similarity threshold, semantic knowledge features corresponding to candidate topic domains that are higher than the first similarity threshold are extracted to obtain knowledge fragments.

[0117] In this embodiment of the application, knowledge fragments are obtained by calculating the cosine similarity between the label type features and the candidate topic domain value vectors.

[0118] For example, first determine the value vector corresponding to each candidate topic domain. This value vector is determined by the importance of knowledge items in the candidate topic domain (measured based on business value or usage frequency) and update time (prioritizing the use of the latest data), ensuring that the generated tags are consistent with the latest business needs.

[0119] Then, the first cosine similarity between the features corresponding to the label type and the value vectors (semantic knowledge features) corresponding to each candidate topic domain is calculated.

[0120] Finally, semantic knowledge features corresponding to candidate topic domains with similarity scores above the first similarity threshold are extracted to obtain knowledge fragments.

[0121] According to embodiments of this application, cosine similarity quantifies semantic relevance by comparing the directions of the label type feature vector and the candidate topic domain value vector, which can filter out the knowledge fragments that best match the label type, ensuring that the extracted knowledge fragments are accurate, efficient and reliable.

[0122] Figure 7 The flowchart illustrating the method for automatic labeling based on candidate labels according to an embodiment of this application is shown.

[0123] like Figure 7 As shown, the method for automatic labeling based on candidate labels in this embodiment includes operations S710 to S720.

[0124] When operating the S710, configure the tagging mechanism, which is configured to automatically tag data assets that meet preset conditions.

[0125] When operating the S720, based on candidate tags, the tagging mechanism is used to tag data assets that meet preset conditions.

[0126] In this embodiment of the application, a tagging mechanism is configured to automatically tag data assets.

[0127] After candidate labels are generated, the multimodal label generation model outputs the configured labeling mechanism for use by the next layer of labeling. This labeling mechanism, or labeling rules, defines "what conditions data assets should meet to be labeled with this type of candidate label." Therefore, under this labeling mechanism, data assets that meet the preset conditions are automatically labeled.

[0128] According to the embodiments of this application, the tagging mechanism defines the tags that should be assigned to data assets when they meet specific conditions. Therefore, the tagging mechanism can be used to quickly complete the automatic tagging operation of data assets that meet preset conditions without the need for manual review, thus reducing the subjective bias of manual tagging.

[0129] Based on the above embodiments, the data asset tag generation method may further include the following operations:

[0130] Perform second cosine similarity detection on data tags of the same type; in response to the second cosine similarity being lower than the second similarity threshold, trigger a tag addition suggestion; based on the tag addition suggestion, use a multimodal tag automatic generation model to generate new data tags, and label data assets that meet preset conditions based on the new data tags.

[0131] In this embodiment of the application, after the data assets are tagged, the similarity of data tags of the same type is detected. When a data tag gap is detected, a new tag is generated in a timely manner.

[0132] There are two ways to initiate the creation of new data tags: one is based on a clear data requirement, and the other is based on keyword recommendations related to industry trends. However, both require review by the data tag administrator before submitting the tag creation request. For both cases, a second cosine similarity check can be initiated for similar data tags. When the second cosine similarity is lower than the second similarity threshold, a tag addition suggestion is triggered.

[0133] For example, after the launch of green credit products, the keyword "carbon emissions" is frequently mentioned and recommended to the tag administrator. If a data tag gap is detected, the tag administrator can request the addition of new carbon emission-related data tags.

[0134] After the request is submitted to the multimodal label automatic generation model, a new list of candidate labels is generated. Based on this new list of candidate labels, the corresponding data assets are labeled.

[0135] According to the embodiments of this application, cosine similarity detection of data tags of the same type can promptly identify data tag gaps and generate new data tags for the data tag gaps, ensuring that the tag system keeps pace with the times, covers emerging fields, and realizes full lifecycle management of data tags.

[0136] Based on the above embodiments, the data asset tag generation method may further include the following operations:

[0137] Calculate the activity level of data tags in each business area; in response to activity levels exceeding an activity threshold, identify data tags exceeding the activity threshold as high-activity data tags; based on high-activity data tags, recommend data assets corresponding to the high-activity data tags to users in the corresponding business area.

[0138] In this embodiment of the application, based on the activity level of data tags in each business area, highly popular data assets involving high-activity data tags are recommended to users in the relevant business areas.

[0139] Data tags can be applied to various data asset query platforms, supporting tag selection and tag exploration. The tag selection function allows users to specify multiple data tags from existing data tags to jointly select data assets. For example, users can select data assets with the business tag "credit card customer acquisition," the technology tag "topic aggregation table," the management tag "high quality score," and the operations tag "high popularity in the credit card field," and then initiate data exploration. The tag exploration function supports fuzzy search, outputting similar data tags, and providing the corresponding data assets after selecting a target tag. Therefore, based on user clicks on data tag tracking tables and data asset query audit logs, daily activity levels in business areas can be calculated for data tags.

[0140] This activity level For example, it can be calculated in the following way:

[0141]

[0142] in, λ represents the activity level of the k-th tag category at time t, and λ represents the time decay factor. This represents the activity level of the k-th tag category at time (t-1). Parameters indicating the importance of each influencing factor Indicating influencing factors, such as the number of times users click on data tags daily on each platform, the number of data tags collected by users, and the number of times tagged data assets are queried, etc.

[0143] According to embodiments of this application, prioritizing the recommendation of relevant high-traffic data assets based on data tag activity can help users quickly obtain data highly relevant to current industry trends or business needs, thereby improving decision-making efficiency.

[0144] In addition, when existing data tags no longer meet business needs or have extremely low activity, they will change from an available state to an offline state, which means the tags are retired.

[0145] In this application embodiment, for tag retirement, on the one hand, the data tag retirement is promoted by manually reviewing the tags, and on the other hand, the tag retirement is implemented automatically based on the judgment model.

[0146] The decision model could be, for example:

[0147]

[0148] Where a and b represent the thresholds for limiting the retirement of data tags, When the activity level of a tag drops significantly and is low, it will be decommissioned.

[0149] Based on the above-described data asset tag generation method, this application also provides a data asset tag generation apparatus. The following will combine... Figure 8 The device is described in detail.

[0150] Figure 8 The diagram illustrates the structure of a data asset tag generation apparatus according to an embodiment of this application.

[0151] like Figure 8 As shown, the data asset tag generation device 800 of this embodiment includes: an acquisition module 810, a determination module 820, a first generation module 830, a second generation module 840, and a tagging module 850.

[0152] The acquisition module 810 is used to acquire the data tag requirements of the user for managing data assets. In one embodiment, the acquisition module 810 can be used to perform the operation S210 described above, which will not be repeated here.

[0153] Input the data labeling requirements into the pre-built multimodal labels to automatically generate the model and execute the following modules:

[0154] The determining module 820 is used to determine the tag type corresponding to the data tag requirements based on the data tag requirements. In one embodiment, the determining module 820 can be used to perform the operation S220 described above, which will not be repeated here.

[0155] The first generation module 830 is used to extract knowledge fragments corresponding to the tag types from a pre-built knowledge base, and generate contextual prompts based on the knowledge fragments. The knowledge base is pre-divided into multiple types of topic domains, and each type of topic domain includes multiple knowledge fragments corresponding to each tag type. In one embodiment, the first generation module 830 can be used to perform the operation S230 described above, which will not be repeated here.

[0156] The second generation module 840 is used to generate candidate tags based on knowledge fragments and contextual prompts. In one embodiment, the second generation module 840 can be used to perform the operation S240 described above, which will not be repeated here.

[0157] The tagging module 850 is used to tag data assets that meet preset conditions based on candidate tags. In one embodiment, the tagging module 850 can be used to perform the operation S250 described above, which will not be repeated here.

[0158] According to an embodiment of this application, the first generation module 830 includes: a raw data acquisition module, a preprocessing module, and a knowledge graph generation module.

[0159] The raw data acquisition module is used to acquire raw data information.

[0160] The preprocessing module is used to preprocess the raw data and divide the preprocessed raw data into multiple subject domain knowledge bases. The knowledge bases are used to provide background knowledge for generating candidate tags.

[0161] The knowledge graph generation module is used to generate a knowledge graph with multiple topic domains and a corresponding vector index based on a knowledge base with multiple topic domains using a knowledge fusion model.

[0162] According to embodiments of this application, the knowledge graph generation module includes: a graph relationship feature extraction module, a key information feature extraction module, a word frequency feature extraction module, a resource attribute and relationship feature extraction module, and a fusion module.

[0163] The graph relation feature extraction module is used to extract graph relation features from graph data based on graph data in a knowledge base using graph neural networks.

[0164] The key information feature extraction module is used to extract key information features from complex text data based on the knowledge base using the first text classification model.

[0165] The word frequency feature extraction module is used to extract word frequency features from simple text data based on knowledge base using a second text classification model.

[0166] The resource attribute and relationship feature extraction module is used to extract the resource attributes and relationship features of each metadata in the structured data based on the structured data in the knowledge base using a standard data model.

[0167] The fusion module is used to fuse graph relationship features, key information features, word frequency features, resource attributes and relationship features to generate a knowledge graph corresponding to each subject domain. In the knowledge graph, nodes represent entity features, and edges represent relationship features between entity features.

[0168] According to an embodiment of this application, the first generation module 830 further includes: a subgraph extraction module, a candidate topic domain determination module, and a similarity calculation module.

[0169] The subgraph extraction module is used to determine the candidate range of the knowledge base based on the tag type, and extract the subgraphs in the knowledge graph that correspond to the tag type within the candidate range.

[0170] The candidate topic domain determination module is used to determine candidate topic domains for a subgraph based on vector indices.

[0171] The similarity calculation module is used to calculate the similarity between the label type and each candidate topic domain within the subgraph, and extract the candidate topic domains that are higher than the first similarity threshold to obtain knowledge fragments.

[0172] According to an embodiment of this application, the similarity calculation module includes: a value vector determination module, a first cosine similarity calculation module, and a semantic knowledge feature extraction module.

[0173] The value vector determination module is used to determine the value vector corresponding to each candidate topic domain based on the importance of the candidate topic domain and the time update parameters. The value vector represents the semantic knowledge features of each candidate topic domain.

[0174] The first cosine similarity calculation module is used to calculate the first cosine similarity between the features corresponding to the label type and the semantic knowledge features corresponding to each candidate topic domain.

[0175] The semantic knowledge feature extraction module is used to extract semantic knowledge features corresponding to candidate topic domains that have a first cosine similarity higher than the first similarity threshold in response to the first cosine similarity being higher than the first similarity threshold, thereby obtaining knowledge fragments.

[0176] According to an embodiment of this application, the marking model 850 includes a configuration module and a marking submodule.

[0177] The configuration module is used to configure the tagging mechanism, which is configured to automatically tag data assets that meet preset conditions.

[0178] The tagging submodule is used to tag data assets that meet preset conditions based on candidate tags using a tagging mechanism.

[0179] According to an embodiment of this application, the data asset tag generation device 800 further includes: a detection module, a triggering module, and a new data tag generation module.

[0180] The detection module is used to perform second cosine similarity detection on data tags of the same type.

[0181] The trigger module is used to trigger a new label suggestion in response to the second cosine similarity being lower than the second similarity threshold.

[0182] The new data tag generation module is used to generate new data tags based on tag addition suggestions, using a multimodal tag generation model, and then tag data assets that meet preset conditions based on the new data tags.

[0183] According to an embodiment of this application, the data asset tag generation device 800 further includes: an activity calculation module, a high-activity data tag determination module, and a recommendation module.

[0184] The activity calculation module is used to calculate the activity level of data tags in each business area.

[0185] The high-activity data tag determination module is used to determine data tags with activity levels above an activity threshold as high-activity data tags in response to such activity levels.

[0186] The recommendation module is used to recommend data assets corresponding to high-activity data tags to users in the corresponding business areas.

[0187] According to embodiments of this application, any multiple modules among the acquisition module 810, determination module 820, first generation module 830, second generation module 840, and marking module 850 can be combined into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of this application, at least one of the acquisition module 810, determination module 820, first generation module 830, second generation module 840, and marking module 850 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), programmable logic array (PLA), system-on-a-chip, system-on-a-substrate, system-on-package, application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging the circuitry, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in a suitable combination of any of these. Alternatively, at least one of the acquisition module 810, determination module 820, first generation module 830, second generation module 840 and marking module 850 can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.

[0188] Figure 9 A block diagram schematically illustrates an electronic device suitable for implementing a data asset tag generation method according to an embodiment of this application.

[0189] like Figure 9 As shown, an electronic device 900 according to an embodiment of this application includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage portion 908 into a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of this application.

[0190] RAM 903 stores various programs and data required for the operation of electronic device 900. Processor 901, ROM 902, and RAM 903 are interconnected via bus 904. Processor 901 executes various operations of the method flow according to embodiments of this application by executing programs in ROM 902 and / or RAM 903. It should be noted that the programs may also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 may also execute various operations of the method flow according to embodiments of this application by executing programs stored in said one or more memories.

[0191] According to embodiments of this application, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to a bus 904. The electronic device 900 may also include one or more of the following components connected to the input / output (I / O) interface 905: an input section 906 including a keyboard, mouse, etc.; an output section 907 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN card, modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the input / output (I / O) interface 905 as needed. A removable medium 911, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 910 as needed so that computer programs read from it can be installed into the storage section 908 as needed.

[0192] This application also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0193] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this application, the computer-readable storage medium may include ROM 902 and / or RAM 903 and / or one or more memories other than ROM 902 and RAM 903 described above.

[0194] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to cause the computer system to implement the methods provided in the embodiments of this application.

[0195] When the computer program is executed by the processor 901, it performs the functions defined in the system / apparatus of this application embodiment. According to the embodiments of this application, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0196] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via the communication section 909, and / or installed from a removable medium 911. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0197] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, it performs the functions defined in the system of this application embodiment. According to the embodiments of this application, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0198] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0199] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0200] Those skilled in the art will understand that the features described in the various embodiments of this application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments of this application can be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

Claims

1. A method for generating data asset tags, characterized in that, The method includes: Obtain user data tag requirements for managing data assets; Input the data label requirements into the pre-built multimodal label automatic model generation and perform the following operations: Based on the data tagging requirements, determine the tag type corresponding to the data tagging requirements; According to the tag type, knowledge fragments corresponding to the tag type are extracted from a pre-built knowledge base, and contextual prompts are generated based on the knowledge fragments. The knowledge base is pre-divided into multiple types of topic domains, and each type of topic domain includes multiple knowledge fragments corresponding to each tag type. Based on the knowledge fragment and the contextual prompts, candidate tags are generated; Based on the candidate tags, data assets that meet the preset conditions are tagged.

2. The method according to claim 1, characterized in that, The automatic multimodal label generation model is built upon a large model, and its structure includes: The tag type layer is configured to determine the tag type corresponding to the data tag requirements based on the data tag requirements. The knowledge fragment extraction layer is configured to extract knowledge fragments corresponding to the tag type from a pre-built knowledge base, and generate contextual prompts based on the knowledge fragments. The tag generation layer is configured to generate candidate tags based on the knowledge fragments and the contextual prompts; The data asset labeling layer is configured to label data assets that meet preset conditions based on the candidate labels.

3. The method according to claim 1, characterized in that, The knowledge base is pre-built, including: Obtain raw data information; The raw data information is preprocessed, and the preprocessed raw data information is divided into multiple topic domain knowledge bases, wherein the knowledge bases are used to provide background knowledge for generating the candidate tags; Based on the knowledge base of the multiple subject domains, a knowledge graph with multiple subject domains and a vector index corresponding to the knowledge graph are generated using a knowledge fusion model.

4. The method according to claim 3, characterized in that, The step of generating a knowledge graph with multiple topic domains using a knowledge fusion model based on the knowledge base of the multiple topics includes: Based on the graph data in the knowledge base, graph neural networks are used to extract graph relationship features from the graph data; Based on the complex text data in the knowledge base, the key information features in the complex text data are extracted using the first text classification model; Based on the simple text data in the knowledge base, the word frequency features in the simple text data are extracted using a second text classification model; Based on the structured data in the knowledge base, the resource attributes and relationship features of each metadata in the structured data are extracted using a standard data model; By integrating the graph relationship features, the key information features, the word frequency features, and the resource attribute and relationship features, a knowledge graph corresponding to each subject domain is generated.

5. The method according to claim 3, characterized in that, The step of extracting knowledge fragments corresponding to the tag type from a pre-built knowledge base includes: Based on the tag type, determine the candidate range of the knowledge base, and within the candidate range, extract the subgraph in the knowledge graph corresponding to the tag type; Based on the vector index, the candidate topic domains of the subgraph are determined; Calculate the similarity between the label type and each candidate topic domain within the subgraph, and extract the candidate topic domains that are higher than the first similarity threshold to obtain knowledge fragments.

6. The method according to claim 5, characterized in that, The calculation of the similarity between the label type and each candidate topic domain within the subgraph, and the extraction of candidate topic domains with similarity values ​​exceeding a first similarity threshold, yields knowledge fragments including: Based on the importance of the candidate topic domains and the time update parameters, a value vector corresponding to each candidate topic domain is determined, wherein the value vector represents the semantic knowledge features of each candidate topic domain. Calculate the first cosine similarity between the features corresponding to the label type and the semantic knowledge features corresponding to each candidate topic domain; In response to the first cosine similarity being higher than the first similarity threshold, semantic knowledge features corresponding to candidate topic domains that are higher than the first similarity threshold are extracted to obtain knowledge fragments.

7. The method according to claim 1, characterized in that, The step of tagging data assets that meet preset conditions based on the candidate tags includes: Configure a tagging mechanism, wherein the tagging mechanism is configured to automatically tag data assets that meet preset conditions; Based on the candidate tags, the tagging mechanism is used to tag data assets that meet preset conditions.

8. The method according to claim 1, characterized in that, The method further includes: Perform second cosine similarity detection on data labels of the same type; In response to the second cosine similarity being lower than the second similarity threshold, a new label suggestion is triggered; Based on the proposed new tags, new data tags are generated using the multimodal tag automatic generation model, and data assets that meet preset conditions are tagged based on the new data tags.

9. The method according to claim 1, characterized in that, The method further includes: Calculate the activity level of data tags in each business area; In response to the activity level being higher than an activity threshold, data tags that are higher than the activity threshold are determined to be high-activity data tags; Based on the high-activity data tags, data assets corresponding to the high-activity data tags are recommended to users in the corresponding business areas.

10. A data asset tag generation device, characterized in that, The device includes: The acquisition module is used to acquire the data tag requirements of users for managing data assets; input the data tag requirements into a pre-built multimodal tag automatic generation model, and execute the following modules: The determination module is used to determine the tag type corresponding to the data tag requirements based on the data tag requirements; The first generation module is used to extract knowledge fragments corresponding to the tag type from a pre-built knowledge base according to the tag type, and generate context prompt words based on the knowledge fragments. The knowledge base is pre-divided into multiple types of topic domains, and each type of topic domain includes multiple knowledge fragments corresponding to each tag type. The second generation module is used to generate candidate tags based on the knowledge fragment and the contextual prompts; The tagging module is used to tag data assets that meet preset conditions based on the candidate tags.

11. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 9.

12. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 9.

Images