Metadata upstream and downstream intelligent analysis method and equipment based on large model and medium

Abstract

The invention discloses a metadata upstream and downstream intelligent analysis method and device based on a large model and a medium, and relates to the field of data processing analysis, and the method comprises the steps: obtaining a to-be-analyzed data entity, and carrying out upstream and downstream dependency relationship conversion on the to-be-analyzed data entity to obtain dependency relationship directed data; acquiring natural language input data, and performing multiple rounds of dialogue analysis on the natural language input data to determine a natural language recognition strategy; based on the dependency directed data and a natural language recognition strategy, through graph retrieval enhancement generation, determining data influence tracking data; according to the data influence tracking data, obtaining a data influence analysis report through state change influence parameter evaluation; and performing data state influence evaluation on the data influence analysis report to determine a data influence visual data set. Through the method, the technical problems of low data influence visualization precision and low intelligent degree in the prior art are solved.

Description

Technical Field

[0001] The present application relates to the field of data processing and analysis, and in particular to a method, device and medium for intelligent upstream and downstream analysis of metadata based on a large model. Background Art

[0002] With the rapid development of big data and artificial intelligence technologies, enterprises and institutions are facing more and more challenges in data management and analysis. Especially in complex data relationship tracking and impact analysis scenarios, traditional tools and methods are often unable to meet the processing needs of dynamic, multi-source and multi-dimensional data.

[0003] Traditional relational databases and data analysis tools often lack sufficient expressiveness when faced with highly complex data relationships, and it is difficult to accurately reveal the deep-level connections between data entities and their upstream and downstream, resulting in inaccurate and incomplete analysis results. Existing data impact visualization methods are not intelligent enough in terms of user interaction experience, and the accuracy of data impact visualization is still insufficient. Summary of the invention

[0004] The embodiments of the present application provide a method, device and medium for intelligent upstream and downstream analysis of metadata based on a large model, which solves the technical problems of low visualization accuracy and low intelligence of data impact in the prior art.

[0005] In a first aspect, an embodiment of the present application provides a metadata upstream and downstream intelligent analysis method based on a large model, characterized in that the method includes: obtaining data entities to be analyzed, and converting upstream and downstream dependencies of the data entities to be analyzed to obtain dependency directed data; obtaining natural language input data, and performing multiple rounds of dialogue analysis on the natural language input data to determine the natural language recognition strategy; based on the dependency directed data and the natural language recognition strategy, determining data impact tracking data through graph retrieval enhancement generation; according to the data impact tracking data, obtaining a data impact analysis report through state change impact parameter evaluation; performing data state impact evaluation on the data impact analysis report to determine a data impact visualization data group.

[0006] In one implementation of the present application, the upstream and downstream dependencies of the data entity to be analyzed are transformed to obtain dependency directed data, specifically including: based on the data entity to be analyzed, determining the key query fields, and based on the key query fields, performing data query on the preset metadata warehouse to obtain data table data; wherein the data table data includes: data table, data table attributes; based on the data table data and the key query fields, determining the data dependency through the construction of a sparse directed graph; based on the data dependency, obtaining dependency directed data.

[0007] In one implementation of the present application, a multi-round dialogue analysis is performed on natural language input data to determine a natural language recognition strategy, specifically including: performing context association analysis on the natural language input data to obtain dialogue association parameters; based on the dialogue association parameters, performing dialogue recognition optimization on the natural language input data, and performing historical dialogue recognition iteration on the context dialogue after dialogue recognition optimization to determine a static multi-round dialogue optimization strategy; performing feedback point analysis on the continuously updated dialogue in the natural language input data to obtain a dynamic multi-round dialogue optimization strategy; and determining a natural language recognition strategy based on the static multi-round dialogue optimization strategy and the dynamic multi-round dialogue optimization strategy.

[0008] In one implementation of the present application, data impact tracking data is determined based on dependency directed data through data impact path generation, specifically including: based on a natural language recognition strategy, determining retrieval key fields, and based on the dependency directed data and the retrieval key fields, determining field dependencies through entity data relationship analysis; based on the field dependencies, determining graph node groups, and performing graph structure reasoning on the graph node groups to obtain data impact tracking data; wherein, the data impact tracking data includes: data lineage analysis data, data impact analysis data of state changes.

[0009] In one implementation of the present application, a data impact analysis report is obtained through state change impact parameter evaluation based on data impact tracking data, specifically including: determining data impact evaluation parameters based on the data impact tracking data, and performing parameter impact threshold judgment on the data impact evaluation parameters to obtain a parameter impact sequence; performing field change matching on the parameter impact sequence to determine a field change impact degree sequence; and performing structured processing on the field change impact degree sequence to obtain a data impact analysis report.

[0010] In one implementation of the present application, a data status impact assessment is performed on the data impact analysis report to determine a data impact visualization data group, specifically including: based on the data impact analysis report, determining a visualization format by matching the impact data type; performing a data table impact assessment on the impact analysis report to obtain first impact status data; performing a downstream system impact assessment on the impact analysis report to obtain second impact status data; and determining a data impact visualization data group by integrating the visualization format, the first impact status data, and the second impact status data through an export format.

[0011] In one implementation of the present application, after determining the data impact visualization data group by exporting the format integration according to the visualization format, the first impact status data, and the second impact status data, the method further includes: inputting the data impact visualization data group into a preset impact prediction model to obtain data impact prediction data; performing a prediction status evaluation on the data impact prediction data to determine data health parameters; and determining a data health alarm report through data health status judgment based on the data health parameters.

[0012] In one implementation of the present application, after performing a data status impact assessment on the data impact analysis report to determine the data impact visualization data group, the method also includes: determining user demand satisfaction through multiple rounds of dialogue interaction based on the visualization data group; performing a refined call configuration on the visualization data group to obtain a refined query report; and performing an association adjustment analysis on the refined query report to determine a downstream data adjustment strategy.

[0013] In the second aspect, an embodiment of the present application also provides a metadata upstream and downstream intelligent analysis method device based on a large model, characterized in that the device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can: obtain data entities to be analyzed, and transform the data entities to be analyzed into upstream and downstream dependencies to obtain dependency directed data; obtain natural language input data, and perform multiple rounds of dialogue analysis on the natural language input data to determine a natural language recognition strategy; based on the dependency directed data and the natural language recognition strategy, determine data impact tracking data through graph retrieval enhancement generation; based on the data impact tracking data, obtain a data impact analysis report through state change impact parameter evaluation; perform data state impact evaluation on the data impact analysis report to determine a data impact visualization data group.

[0014] In the third aspect, the embodiment of the present application also provides a non-volatile computer storage medium for a large model-based metadata upstream and downstream intelligent analysis method, which stores computer executable instructions, characterized in that the computer executable instructions are set to: obtain the data entity to be analyzed, and transform the upstream and downstream dependencies of the data entity to be analyzed to obtain dependency directed data; obtain natural language input data, and perform multiple rounds of dialogue analysis on the natural language input data to determine the natural language recognition strategy; based on the dependency directed data and the natural language recognition strategy, determine the data impact tracking data through graph retrieval enhancement generation; based on the data impact tracking data, obtain a data impact analysis report through state change impact parameter evaluation; perform data state impact evaluation on the data impact analysis report to determine the data impact visualization data group.

[0015] The embodiments of the present application provide a metadata upstream and downstream intelligent analysis method, device and medium based on a large model. Through metadata modeling, data tracking, multi-round dialogue intelligent interaction and data dependency analysis and display, the technical problems of low data impact visualization accuracy and low intelligence in the prior art are solved, and the impact of changes in a key field in business-related metadata on upstream and downstream data is analyzed, which improves the accuracy and efficiency of data lineage and impact analysis of metadata and optimizes the user's actual operation experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 A flowchart of a metadata upstream and downstream intelligent analysis method based on a large model provided in an embodiment of the present application; Figure 2 A schematic diagram of the internal structure of a metadata upstream and downstream intelligent analysis device based on a large model provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only 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 making creative work are within the scope of protection of the present application.

[0018] The embodiments of the present application provide a metadata upstream and downstream intelligent analysis method, device and medium based on a large model. Through metadata modeling, data tracking, multi-round dialogue intelligent interaction and data dependency analysis and display, the technical problems of low data impact visualization accuracy and low intelligence in the prior art are solved, and the impact of changes in a key field in business-related metadata on upstream and downstream data is analyzed, which improves the accuracy and efficiency of data lineage and impact analysis of metadata and optimizes the user's actual operation experience.

[0019] The technical solution proposed in the embodiments of the present application is described in detail below with reference to the accompanying drawings.

[0020] Figure 1 The embodiment of the present application provides a metadata upstream and downstream intelligent analysis method based on a large model. Figure 1 As shown, the embodiment of the present application provides a metadata upstream and downstream intelligent analysis method based on a large model, which specifically includes the following steps: Step 101: Acquire the data entity to be analyzed, and transform the upstream and downstream dependency relationships of the data entity to be analyzed to obtain dependency directed data.

[0021] Specifically, it includes: based on the data entity to be analyzed, determining the key query fields, and based on the key query fields, performing data query on the preset metadata warehouse to obtain data table data; wherein the data table data includes: data table, data table attributes; based on the data table data and the key query fields, determining the data dependency through sparse directed graph construction; based on the data dependency, obtaining dependency directed data.

[0022] This application achieves metadata modeling by acquiring the data entities to be analyzed and transforming the upstream and downstream dependencies of the data entities to be analyzed to obtain dependency directed data, thereby providing intuitive dependency representation data for subsequent metadata lineage tracking and data analysis.

[0023] In the examples of the present application, detailed explanation is given by the following Example 1.

[0024] Example 1: Data lineage refers to the naturally formed relationship between data during the process of data generation, ETL processing, processing, fusion, circulation and final extinction, which represents the source, conversion process, flow and final use of the data.

[0025] E-commerce platform A plans to make structural adjustments to its order database, changing the "Payment Status" field in the order table from a Boolean value (paid / unpaid) to a multi-status value (pending payment, paying, paid, payment failed, etc.). This change may affect multiple systems and reports that rely on this field.

[0026] Regarding determining the key query field, the key query field in this embodiment is set to "payment status", and the data table where the "payment status" field is located and its attribute information are obtained by querying the metadata warehouse.

[0027] First, identify and define all data entities, including databases, data tables, fields, files, data pipelines, application systems, etc. For each data entity type (such as databases, data tables, fields, etc.), establish a comprehensive metadata catalog through classification and hierarchical management. Add detailed attribute information to each data entity, including name, type, data format, system, business meaning, etc., to improve the description of metadata.

[0028] Then, define various relationships between data entities, including dependency, inclusion, and derivative relationships. By defining the relationship node structure, add a pointer to the relationship list in the data entity structure, create a relationship list for each data entity, and clarify its association with other entities. Subsequently, use methods such as the entity-relationship model (ER model) to model these entities and their relationships, and systematically represent the data structure.

[0029] Finally, based on the above data entities and relationships, the data entities are mapped to nodes in the graph, the dependencies between entities are mapped to directed edges, and relevant attribute information is added to the nodes and edges. By mapping nodes to directed edges, a sparse directed graph is constructed and stored, so that the data entities and their complex dependencies can be displayed intuitively.

[0030] In practical applications, the above method can be integrated through a metadata modeling module to achieve the technical effects achieved by the above method.

[0031] Step 102: Obtain natural language input data, and perform multiple rounds of dialogue analysis on the natural language input data to determine a natural language recognition strategy.

[0032] Specifically, it includes: performing context association analysis on natural language input data to obtain dialogue association parameters; performing dialogue recognition optimization on natural language input data based on dialogue association parameters, and iterating historical dialogue recognition on the context dialogue after dialogue recognition optimization to determine a static multi-round dialogue optimization strategy; performing feedback point analysis on continuously updated dialogues in natural language input data to obtain a dynamic multi-round dialogue optimization strategy; and determining a natural language recognition strategy based on the static multi-round dialogue optimization strategy and the dynamic multi-round dialogue optimization strategy.

[0033] This application obtains natural language input data and performs multi-round conversation analysis on the natural language input data to determine the natural language recognition strategy, thereby achieving semantic intelligent recognition during use and improving the accuracy of data impact analysis and data lineage recognition in multi-round conversations, thereby optimizing the user's actual operating experience.

[0034] In the examples of the present application, detailed explanation is given by the following Example 2.

[0035] Example 2: Multi-round dialogue requires the introduction of a natural language interaction interface that can remember and utilize the user's historical interaction information to simplify the execution of multi-step queries.

[0036] In order to obtain the impact of field changes on the overall data, the user enters a query statement such as "I want to know which data tables, processing procedures and systems will be affected after the 'payment status' field is modified."

[0037] First, through natural language analysis, the user's natural language input is received, and the query intent is identified through intelligent analysis technology, supporting voice and text input. The user input is analyzed through a multi-round dialogue interaction module to identify the core intent of the query, that is, the user needs to understand which downstream data or preset systems will be affected after the "payment status" field is changed.

[0038] Then, the user's historical query content is recorded and queried, and the analysis results are optimized according to the context in multiple rounds of conversations to improve the accuracy of multi-step queries. Since it is aimed at static data that has been completed and recorded, the contextual conversation after conversation recognition optimization is iterated on the historical conversation recognition to further improve the accuracy of step queries.

[0039] Finally, by dynamically adjusting the analysis process, the analysis results are gradually refined according to user feedback to ensure that complex tasks are accurately processed. The dialogues input by users during use and the corresponding keywords or key words in the answers to the input dialogues are regarded as feedback points. Through the natural language processing mechanism based on contextual memory, user intentions are identified and complex tasks are decomposed into multiple subtasks.

[0040] It should be noted that during the task parsing process, based on the user's query-related data, the task decomposition algorithm is used to decompose the problem into the following subtasks (based on payment status changes): Query the dependencies of the "payment status" field in the system; Track downstream data and systems affected by the “Payment Status” field; Generate a change impact analysis report.

[0041] The above subtasks are assigned through the task parsing and disassembly module and executed in order of priority. For example, if the priority of the first and second subtasks is 0, it means that they can be executed in parallel, and the priority of the third subtask is 1, which means that it can only be executed after the first two tasks are completed and the results are obtained. At this point, the system can continue to have multiple rounds of dialogue with the user based on context memory to ensure that the parsed tasks meet the user's expectations.

[0042] Feedback the analysis results to users in the form of natural language. Utilize the natural language generation capability of the large model and combine it with the context memory function to generate easy-to-understand text descriptions.

[0043] The interactive system answers: "The upstream dependencies of data table A include data table B and data table C. Modifying field X may affect fields Y and Z in data table D." Because it supports multiple rounds of dialogue, users can ask further questions, such as "What is the extent of the impact of data table D?"

[0044] The interactive system will continue to provide detailed answers based on the previous conversation content. In this way, the entire query and analysis process is simplified by natural language interaction, and users can efficiently obtain the required information without having a professional technical background.

[0045] In practical applications, the above technical solution can be integrated into a preset multi-round dialogue interaction module to achieve the technical effect achieved by the above method.

[0046] Step 103: Based on the dependency directed data and natural language recognition strategy, the data impact tracking data is determined by generating the data impact path.

[0047] Specifically, it includes: determining the key retrieval fields based on the natural language recognition strategy, and determining the field dependencies through entity data relationship analysis based on the dependency directed data and the key retrieval fields; determining the graph node group based on the field dependencies, and performing graph structure reasoning on the graph node group to obtain data impact tracking data; wherein, the data impact tracking data includes: data lineage analysis data, data impact analysis data of state changes.

[0048] This application is based on dependency-related directed data, and through graph retrieval enhancement generation, it determines the data impact tracking data, realizes the construction of data lineage and the impact of a field change on the data.

[0049] In the examples of the present application, this is explained in detail through the following Example 3.

[0050] Example 3: GraphRAG (Graph-based Retrieval-Augmented Generation) combines graph reasoning and retrieval augmented generation (RAG) methods to automatically capture complex dependency paths between data entities.

[0051] By searching and finally obtaining the closest graph nodes, the graph node group is determined. The upstream and downstream relationships of data entities are determined through graph structure reasoning. Starting from the target data entity node, the directed edges are traversed in reverse to generate the path and results of the upstream dependency relationship, and the data impact tracking data is obtained.

[0052] Track the upstream and downstream dependencies of the "payment status" field, traverse forward along the directed edges, identify the affected data entities downstream, and determine which data flow paths are most likely to be affected by changes to the field through graph structure reasoning and retrieval, that is, data impact tracking data.

[0053] In actual application, the above method can be integrated into the preset data lineage analysis and impact analysis modules to achieve the corresponding technical effects.

[0054] Step 104: According to the data impact tracking data, the data impact analysis report is obtained by evaluating the state change impact parameters.

[0055] Specifically, it includes: determining data impact assessment parameters based on data impact tracking data, and performing parameter impact threshold judgment on the data impact assessment parameters to obtain a parameter impact sequence; performing field change matching on the parameter impact sequence to determine the field change impact degree sequence; and performing structured processing on the field change impact degree sequence to obtain a data impact analysis report.

[0056] This application obtains a data impact analysis report based on data impact tracking data and by evaluating the impact parameters of state changes, thereby achieving a quantitative evaluation of the impact of field changes on upstream and downstream data and improving the accuracy and efficiency of data impact analysis.

[0057] In the examples of the present application, a detailed explanation is given by the following Example 4.

[0058] Example 4: Through GraphRAG dependency graph visualization, when querying how the "payment status" field affects downstream data flow and processing, the relationship dependency graph generated by GraphRAG will show the relationship between each table and "payment status".

[0059] The relationships of "Payment Status" may include: the payment report system relies on this field to generate daily payment reports, the refund processing system relies on this field to trigger automatic refund operations, and the financial system relies on this field for revenue calculation and reporting, etc.

[0060] By judging the parameter impact threshold, sorting the impact parameters within and outside the threshold range can improve the clarity of the data in the analysis report, and then matching the corresponding changed fields.

[0061] The analysis report includes but is not limited to: a list of affected downstream systems and data processing processes, problems or system failures that may be caused by the change, and recommended solutions to these problems.

[0062] Step 105: Perform a data status impact assessment on the data impact analysis report to determine a data impact visualization data group.

[0063] Specifically, it includes: based on the data impact analysis report, determining the visualization format by matching the impact data type; performing a data table impact assessment based on the impact analysis report to obtain first impact status data; performing a downstream system impact assessment based on the impact analysis report to obtain second impact status data; and determining the data impact visualization data group through export format integration based on the visualization format, the first impact status data, and the second impact status data.

[0064] After determining the data impact visualization data group by exporting the format integration according to the visualization format, the first impact status data, and the second impact status data, the method further includes: inputting the data impact visualization data group into a preset impact prediction model to obtain data impact prediction data; performing a prediction status evaluation on the data impact prediction data to determine data health parameters; and determining a data health alarm report through data health status judgment based on the data health parameters.

[0065] After performing a data status impact assessment on the data impact analysis report to determine the data impact visualization data group, the method also includes: determining user demand satisfaction based on the visualization data group through multiple rounds of dialogue interaction; refining the call configuration of the visualization data group to obtain a refined query report; and performing a correlation adjustment analysis on the refined query report to determine a downstream data adjustment strategy.

[0066] This application conducts a data status impact assessment on the data impact analysis report to determine the data impact visualization data group, and realizes the analysis of the impact of changes to a key field in business-related metadata on upstream and downstream data, thereby improving the accuracy and efficiency of data lineage and impact analysis of metadata.

[0067] The above is an embodiment of the method proposed in this application. Based on the same inventive concept, the embodiment of this application also provides a metadata upstream and downstream intelligent analysis device based on a large model, and its structure is as follows: Figure 2 shown.

[0068] Figure 2 The internal structure diagram of a metadata upstream and downstream intelligent analysis device based on a large model provided in the embodiment of the present application. Figure 2 As shown, the device includes: at least one processor 201; and, a memory 202 communicatively connected to the at least one processor; The memory 202 stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor 201 to enable at least one processor 201 to: Acquire the data entities to be analyzed, and transform the upstream and downstream dependencies of the data entities to be analyzed to obtain dependency directed data; acquire natural language input data, and perform multiple rounds of dialogue analysis on the natural language input data to determine the natural language recognition strategy; based on the dependency directed data and the natural language recognition strategy, enhance the generation through graph retrieval to determine the data impact tracking data; based on the data impact tracking data, obtain the data impact analysis report through the state change impact parameter evaluation; perform data state impact evaluation on the data impact analysis report to determine the data impact visualization data group.

[0069] Some embodiments of the present application provide corresponding Figure 1 A non-volatile computer storage medium for metadata upstream and downstream intelligent analysis based on a large model, storing computer executable instructions, wherein the computer executable instructions are set as: Acquire the data entities to be analyzed, and transform the upstream and downstream dependencies of the data entities to be analyzed to obtain dependency directed data; acquire natural language input data, and perform multiple rounds of dialogue analysis on the natural language input data to determine the natural language recognition strategy; based on the dependency directed data and the natural language recognition strategy, enhance the generation through graph retrieval to determine the data impact tracking data; based on the data impact tracking data, obtain the data impact analysis report through the state change impact parameter evaluation; perform data state impact evaluation on the data impact analysis report to determine the data impact visualization data group.

[0070] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the IoT device and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0071] The system and medium provided in the embodiments of the present application correspond one-to-one to the method. Therefore, the system and medium also have similar beneficial technical effects to the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the system and medium will not be repeated here.

[0072] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0073] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0074] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0075] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0076] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0077] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0078] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0079] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0080] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A metadata upstream and downstream intelligent analysis method based on a large model, characterized in that: The method comprises: Acquire a data entity to be analyzed, and transform the upstream and downstream dependency relationships of the data entity to be analyzed to obtain dependency-oriented data; Acquire natural language input data, and perform multiple rounds of dialogue analysis on the natural language input data to determine a natural language recognition strategy; Based on the dependency directed data and the natural language recognition strategy, determining data impact tracking data through data impact path generation; According to the data impact tracking data, a data impact analysis report is obtained by evaluating the state change impact parameters; A data status impact assessment is performed on the data impact analysis report to determine a data impact visualization data group.

2. According to claim 1, a metadata upstream and downstream intelligent analysis method based on a large model is characterized in that: The data entity to be analyzed is transformed into upstream and downstream dependency relationships to obtain dependency directed data, specifically including: Based on the data entity to be analyzed, a key query field is determined, and according to the key query field, a data query is performed on a preset metadata warehouse to obtain data table data; wherein the data table data includes: a data table and data table attributes; According to the data in the data table and the key query fields, a data dependency relationship is determined by constructing a sparse directed graph; Based on the data dependency, dependency directed data is obtained.

3. According to claim 1, a metadata upstream and downstream intelligent analysis method based on a large model is characterized in that: Performing multiple rounds of dialogue analysis on the natural language input data to determine a natural language recognition strategy, specifically including: Performing context association analysis on the natural language input data to obtain a dialogue association parameter; Based on the dialogue correlation parameter, the natural language input data is optimized for dialogue recognition, and the context dialogue after the dialogue recognition optimization is iterated for historical dialogue recognition to determine a static multi-round dialogue optimization strategy; Performing feedback point analysis on the continuously updated dialogue in the natural language input data to obtain a dynamic multi-round dialogue optimization strategy; The natural language recognition strategy is determined according to the static multi-round dialogue optimization strategy and the dynamic multi-round dialogue optimization strategy.

4. According to claim 1, a metadata upstream and downstream intelligent analysis method based on a large model is characterized in that: Based on the dependency directed data and the natural language recognition strategy, data impact tracking data is determined by generating a data impact path, specifically including: Based on the natural language recognition strategy, determine the key search field, and based on the dependency directed data and the key search field, determine the field dependency through entity data relationship analysis; According to the field dependency, a graph node group is determined, and graph structure reasoning is performed on the graph node group to obtain data impact tracking data; wherein the data impact tracking data includes: data lineage analysis data, data impact analysis data of state changes.

5. According to claim 1, a metadata upstream and downstream intelligent analysis method based on a large model is characterized in that: According to the data impact tracking data, a data impact analysis report is obtained by evaluating the state change impact parameters, including: Based on the data impact tracking data, determine the data impact assessment parameters, and perform parameter impact threshold judgment on the data impact assessment parameters to obtain a parameter impact sequence; Performing field change matching on the parameter impact sequence to determine a field change impact degree sequence; The field change impact degree sequence is structured to obtain the data impact analysis report.

6. The method for upstream and downstream intelligent analysis of metadata based on a large model according to claim 1, characterized in that: Performing a data status impact assessment on the data impact analysis report to determine a data impact visualization data group, specifically including: Based on the data impact analysis report, determining a visualization format by matching the impact data types; Performing a data table impact assessment on the data impact analysis report to obtain first impact status data; Performing a downstream system impact assessment on the impact analysis report to obtain second impact status data; The data impact visualization data group is determined based on the visualization format, the first impact status data, and the second impact status data through export format integration.

7. The method for upstream and downstream intelligent analysis of metadata based on a large model according to claim 6, characterized in that: After determining the data impact visualization data group according to the visualization format, the first impact status data, and the second impact status data by exporting the format integration, the method further includes: Inputting the data impact visualization data set into a preset impact prediction model to obtain data impact prediction data; Performing a forecast status assessment on the data impact forecast data to determine data health parameters; Based on the data health parameters, a data health alarm report is determined through data health status determination.

8. The method for upstream and downstream intelligent analysis of metadata based on a large model according to claim 1, characterized in that: After performing a data status impact assessment on the data impact analysis report to determine a data impact visualization data group, the method further includes: Based on the visualization data group, determining user demand satisfaction through multiple rounds of dialogue interaction; Refining and calling the visualization data group to obtain a refined query report; A correlation adjustment analysis is performed on the refined query report to determine a downstream data adjustment strategy.

9. A metadata upstream and downstream intelligent analysis device based on a large model, characterized in that: The device comprises: at least one processor; and, a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: Acquire a data entity to be analyzed, and transform the upstream and downstream dependency relationships of the data entity to be analyzed to obtain dependency-oriented data; Acquire natural language input data, and perform multiple rounds of dialogue analysis on the natural language input data to determine a natural language recognition strategy; Based on the dependency directed data and the natural language recognition strategy, determining data impact tracking data through data impact path generation; According to the data impact tracking data, a data impact analysis report is obtained by evaluating the state change impact parameters; A data status impact assessment is performed on the data impact analysis report to determine a data impact visualization data group.

10. A non-volatile computer storage medium for metadata upstream and downstream intelligent analysis based on a large model, storing computer executable instructions, characterized in that: The computer executable instructions are configured to: Acquire a data entity to be analyzed, and transform the upstream and downstream dependency relationships of the data entity to be analyzed to obtain dependency-oriented data; Acquire natural language input data, and perform multiple rounds of dialogue analysis on the natural language input data to determine a natural language recognition strategy; Based on the dependency directed data and the natural language recognition strategy, determining data impact tracking data through data impact path generation; According to the data impact tracking data, a data impact analysis report is obtained by evaluating the state change impact parameters; A data status impact assessment is performed on the data impact analysis report to determine a data impact visualization data group.

Images