A data result acquisition method and system based on large model analysis

Abstract

The present application relates to the technical field of data processing, and more particularly to a data result acquisition method and system based on large model analysis, comprising: acquiring a query request input by a user; performing semantic analysis on the query request to obtain a set of query elements; substituting the set of query elements into a preset dynamic knowledge base, performing similarity matching between the set of query elements and the dynamic knowledge base to obtain a query instruction; inputting the query instruction into a preset sentence optimization model, and outputting a query sentence by the sentence optimization model; and based on the query sentence and a preset dynamic metadata management model, performing automatic cross-system data association query to obtain a reply result responding to the user query request.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for obtaining data results based on large model analysis. Background Technology

[0002] In today's information-saturated era, data has become a core driver of development across industries, and users' demand for quick and accurate data results is increasingly urgent. Whether it's an individual user's daily information inquiry or a business user's business data analysis, efficient data acquisition methods are essential.

[0003] Traditional data retrieval methods suffer from low accuracy in semantic understanding when parsing query requests. They typically rely on fixed keyword matching or simple rule engines, failing to delve into the implicit semantics and complex logic within the query request, resulting in incomplete and inaccurate retrieved query elements. This directly impacts the efficiency and accuracy of subsequent data queries. Summary of the Invention

[0004] This invention provides a data result acquisition method and system based on large model analysis that can improve the efficiency and accuracy of result acquisition, and can effectively solve the problems in the background art.

[0005] To achieve the above objectives, in a first aspect, the present invention provides a method for obtaining data results based on large model analysis, comprising:

[0006] Get the query request input by the user;

[0007] Perform semantic parsing on the query request to obtain the query element set;

[0008] Substitute the query element set into the preset dynamic knowledge base, perform similarity matching between the query element set and the dynamic knowledge base, and obtain the query instruction.

[0009] Input the query command into the preset statement optimization model, and the statement optimization model outputs the query statement;

[0010] Based on the query statement and the preset dynamic metadata management model, automatic cross-system data association queries are performed to obtain response results in response to user query requests.

[0011] In conjunction with the first aspect, in one possible design, the query request includes at least one of a text request, a voice request, and an image request.

[0012] In conjunction with the first aspect, in one possible design, the query element set includes query intent, key entities, logical relationships, contextual information, and business terms.

[0013] In conjunction with the first aspect, one possible design involves performing similarity matching between the query element set and a dynamic knowledge base to obtain the query command:

[0014] The query elements are mapped in a dynamic knowledge base to identify the standardized terms and business definitions corresponding to the elements.

[0015] Logical matching is performed using predefined indicator calculation rules and dimension relationships in the knowledge base to determine the required data analysis logic and calculation methods;

[0016] The built-in thesaurus of terms in the knowledge base is used to perform fuzzy matching and error correction on the parts of the query elements that have non-standard or ambiguous expressions.

[0017] The query element set is converted into query instructions by using the mapping results, logical matching results, and term thesaurus query results.

[0018] In conjunction with the first aspect, in one possible design, the query instruction includes the data query range, calculation logic, and correlation dimensions.

[0019] In conjunction with the first aspect, in one possible design, the process of the statement optimization model outputting the query statement includes:

[0020] The query optimization model analyzes the syntax of the query command and examines its basic syntax rules.

[0021] Perform semantic analysis on query commands to obtain their implicit meanings;

[0022] The query commands are processed according to user preferences by combining the user's query history and context information;

[0023] The query optimization model generates the final optimized query statement based on syntactic and semantic analysis and user preference processing results.

[0024] In conjunction with the first aspect, one possible design involves automatically performing cross-system data association queries based on the query statement and a pre-defined dynamic metadata management model to obtain the response results in response to the user's query request:

[0025] Based on the system association information recorded by the dynamic metadata management model, the data sources involved in the query statement are analyzed, the query statement is broken down into sub-query tasks adapted to each system, and routed to the corresponding data interface.

[0026] For data that requires cross-system joint queries, data concatenation is processed according to predefined association rules;

[0027] Call the real-time data interfaces of each system to execute subqueries and obtain the raw data results;

[0028] The raw data returned by multiple systems is standardized to generate a structured integrated result;

[0029] Transform the integrated results into responses that users can understand.

[0030] In conjunction with the first aspect, in one possible design, the response result in response to a user query request comprises structured data tables, visual charts, and natural language analysis conclusions.

[0031] Secondly, the present invention also provides a data result acquisition system based on large model analysis, comprising:

[0032] The query request acquisition module acquires the query request input by the user; the query request includes at least one of text request, voice request, and image request;

[0033] The semantic parsing module performs semantic parsing on the query request to obtain the query element set;

[0034] The similarity matching module substitutes the query element set into the preset dynamic knowledge base, performs similarity matching between the query element set and the dynamic knowledge base, and obtains the query command.

[0035] The query statement optimization module takes the query command as input into the preset statement optimization model, and the statement optimization model outputs the query statement.

[0036] The cross-system data query module automatically correlates and queries data across systems based on the query statement and the preset dynamic metadata management model, and obtains the response results in response to the user's query request.

[0037] In conjunction with the second aspect, in one possible design, the similarity matching module further includes an error correction unit.

[0038] The technical solution of this invention can achieve the following technical effects:

[0039] Breaking through the limitations of traditional fixed keyword matching or simple rule engines, it can deeply understand the implicit semantics and complex logic in query requests; it supports automatic conversion between business terminology and colloquial expressions, with high accuracy in recognizing industry terms and proper nouns, avoiding errors in query element extraction due to misunderstandings of terminology; it supports multimodal input, including text, voice, and images, meeting users' query needs in different scenarios; it can handle complex queries such as multi-dimensional combined queries and trend predictions, no longer limited to simple single-table queries; at the same time, with the help of dynamic knowledge bases and cross-method data queries, it can achieve automatic cross-method data association queries without the need for manual configuration of association relationships, effectively solving the data silo problem; query statement optimization is based on a preset statement optimization model, which can automatically optimize the execution efficiency of query statements and reduce... Queries are time-consuming; moreover, combined with a query result caching mechanism, repeated accesses to the same question can directly return historical results, further shortening response time from the traditional hours or even days to seconds; cross-method data queries are automatically correlated based on a preset dynamic metadata management model, eliminating complex manual processes and significantly improving the efficiency of data query and integration; it lowers the operational threshold for users, reduces reliance on IT department data query support, and allows non-technical personnel to independently complete complex data analysis, reducing user training costs and enterprise labor costs; it has functions such as sensitive data de-identification processing and full lifecycle management of operation logs, recording all elements such as query conditions, result data, and access time, facilitating audit traceability and reducing the risk of sensitive data leakage. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a flowchart of the present invention;

[0042] Figure 2 To obtain the system's structure diagram based on the data results of large model analysis; Detailed Implementation

[0043] This application will now be described with reference to the accompanying drawings.

[0044] like Figure 1 As shown, the present invention provides a method for obtaining data results based on large model analysis, which specifically includes the following steps:

[0045] S1. Obtain the query request input by the user; the query request includes at least one of text request, voice request, and image request;

[0046] S2. Perform semantic parsing on the query request to obtain the query element set;

[0047] S3. Substitute the query element set into the preset dynamic knowledge base, perform similarity matching between the query element set and the dynamic knowledge base, and obtain the query instruction.

[0048] S4. Input the query command into the preset statement optimization model, and the statement optimization model outputs the query statement;

[0049] S5. Based on the query statement and the preset dynamic metadata management model, perform automatic cross-system data association queries to obtain the response results in response to the user's query request.

[0050] This embodiment breaks through the limitations of traditional fixed keyword matching or simple rule engines, enabling a deeper understanding of the implicit semantics and complex logic in query requests; it supports automatic conversion between business terminology and colloquial expressions, with high accuracy in recognizing industry terms and proper nouns, avoiding errors in query element extraction due to misunderstandings of terminology; it supports multimodal input, including text, voice, and images, meeting users' query needs in different scenarios; it can handle complex queries such as multi-dimensional combined queries and trend predictions, no longer limited to simple single-table queries; simultaneously, leveraging a dynamic knowledge base and cross-method data queries, it achieves automatic cross-method data association queries without the need for manual configuration of association relationships, effectively solving the data silo problem; query statement optimization is based on a preset statement optimization model, which can automatically optimize the execution efficiency of query statements. This reduces query time and improves efficiency. Furthermore, combined with a query result caching mechanism, repeated accesses to the same query can directly return historical results, further shortening response time from hours or even days to seconds. Cross-method data queries are automatically linked based on a pre-defined dynamic metadata management model, eliminating complex manual processes and significantly improving the efficiency of data querying and integration. It lowers the operational threshold for users, reduces reliance on IT support for data queries, and allows non-technical personnel to independently complete complex data analysis, reducing user training costs and corporate labor costs. It also features sensitive data anonymization and full lifecycle management of operation logs, recording all elements such as query conditions, result data, and access time, facilitating auditing and traceability, and reducing the risk of sensitive data leakage.

[0051] In some embodiments of the present invention, for step S1, the query request input by the user is obtained;

[0052] The query request includes at least one of text request, voice request, and image request;

[0053] Text Request: Users can directly enter their query statements in text form in the system's query interface; text input is suitable for most users who are familiar with database queries or have specific query needs; the system will preprocess the input text to ensure accurate semantic parsing in subsequent steps.

[0054] Voice Requests: Some users may prefer to interact via voice, so we integrate voice recognition technology. Users can speak their query requests through a voice input device, and the system will convert the speech into text for subsequent processing. This improves the convenience of user input and allows the system to adapt to the needs of more scenarios. The accuracy of voice recognition affects the accurate transmission of query requests, so a high-precision voice recognition algorithm needs to be set up and trained on a large corpus to ensure accurate speech-to-text conversion in various environments.

[0055] Image Requests: In certain scenarios, users may wish to submit query requests by uploading images. Image recognition technology is integrated to parse and extract information such as tables and charts from images. Through optical character recognition and image processing technology, text, numbers, and other information in images are converted into processable text for subsequent semantic analysis and query processing.

[0056] In this embodiment, users can choose the appropriate input method according to their needs, enabling the system to cover a wider range of users. Whether users are accustomed to traditional text queries or prefer voice or image input, they can all obtain a good interactive experience. The support for voice and image requests enhances the system's convenience and accessibility for scenarios where it is inconvenient to use a keyboard, providing a more user-friendly interaction method. Text requests are suitable for users with clear query needs, while voice requests can better serve users who are on the move or cannot focus on keyboard operation. The introduction of image requests allows the system to provide innovative solutions for visual information input. Users can upload information such as charts and forms, and the system can automatically recognize and provide... Extracting text and data from images for further query processing is particularly suitable for scenarios requiring the handling of large amounts of data visualization information. Integrating speech recognition and image recognition technologies allows users to input queries in a more natural way, greatly improving input efficiency. Voice input avoids tedious keyboard operations, while image input solves the problem of extracting information from static content. High-precision speech and image recognition technologies effectively reduce the risk of human input errors and ensure accurate transmission of user intent even in noisy environments. The combination of image processing and optical character recognition technologies enables the system to process non-text data such as graphics and tables, enhancing its adaptability to diverse data scenarios.

[0057] In some embodiments of the present invention, for step S2, semantic parsing is performed on the query request to obtain a set of query elements;

[0058] Before performing semantic parsing on the query request, the query request input by the user is classified and preprocessed;

[0059] When the query request is text, the preprocessing process includes removing stop words, segmenting words, tagging parts of speech, and removing redundant symbols; in addition, spell correction and synonym replacement are also required to ensure the consistency of the input text.

[0060] Voice requests need to be recognized first and converted into text, transforming the voice signal into a text form that can be used for subsequent text processing.

[0061] Image request processing requires the use of computer vision technology; first, important information in the image is extracted through image recognition or object detection algorithms and then converted into text descriptions or labels;

[0062] Perform semantic parsing on the query request to obtain the query element set:

[0063] Use a large language model to classify query requests by intent, determine the purpose of the user's query, and guide the query requests to the correct processing flow.

[0064] Key entities can be identified and extracted from query requests, and through entity relationship extraction technology, the relationships between entities can also be understood.

[0065] For complex logic contained in query requests, the contextual understanding capabilities of large language models are used, combined with predefined logic parsing rules, to convert the logic described in natural language into machine-processable logical expressions.

[0066] The results of intent recognition, entity extraction, and complex logical expression parsing are integrated to form a structured set of query elements;

[0067] The query element set contains all the key information needed to construct a query instruction, including query intent, key entities, logical relationships, contextual information, and business terms.

[0068] In this embodiment, by classifying, preprocessing, and semantically parsing query requests, the user's true intent and key information can be effectively extracted, ensuring that subsequent processing steps are more accurate and efficient. This step not only handles text requests, achieving comprehensive support for different types of input; it performs unified semantic parsing and processing on different forms of query requests such as text, voice, and images, ensuring the stability and consistency of the system under various input methods and avoiding processing deviations caused by different input methods. By utilizing large language models for intent classification, entity extraction, and logical expression parsing, the details of query requests can be deeply understood, key information in the query can be identified and associated, and the depth of query understanding can be improved. Through semantic parsing and contextual understanding of query requests, complex logical relationships in the query can be better identified, ensuring that the final generated query element set can accurately express the user's needs, improving the relevance and accuracy of the query results. By converting natural language into a machine-processable structured query element set, it facilitates subsequent automated processing, reduces manual intervention, and improves the system's intelligence level.

[0069] In some embodiments of the present invention, for step S3, the query element set is substituted into a preset dynamic knowledge base, and the query element set and the dynamic knowledge base are matched for similarity to obtain a query instruction.

[0070] The core components of the dynamic knowledge base include:

[0071] Domain ontology: Automatically extracts core terms and entity relationships from business scenarios to build a structured knowledge system, clarifying the hierarchical relationships and business logic of each element;

[0072] Real-time metadata: Synchronize the database table structure, field meanings, indicator calculation rules and other metadata information of the business system to ensure that the knowledge base is consistent with the actual data storage logic, with an update delay of ≤5 minutes to ensure the timeliness of matching;

[0073] Historical interaction data: This data preserves users' past query elements, matching results, and feedback information.

[0074] The query element set is mapped to the domain ontology in the dynamic knowledge base to identify the standardized terms and business definitions corresponding to the elements;

[0075] For query elements containing complex logic, logical matching is performed using predefined indicator calculation rules and dimensional relationships in the knowledge base to determine the required data analysis logic and calculation methods;

[0076] When the query elements are not expressed in a standardized or ambiguous way, the built-in thesaurus of terms in the knowledge base performs fuzzy matching and error correction to improve the matching error tolerance.

[0077] Through the above similarity matching, the system converts the query element set into a structured query instruction that can be understood by a computer. The query instruction includes a clear data query scope, specific calculation logic, and related dimensions.

[0078] In this embodiment, by performing similarity matching between the query element set and the domain ontology, real-time metadata, and historical interaction data in the dynamic knowledge base, the query elements can be accurately identified and transformed into standardized terms and business definitions. This ensures that the query instructions accurately reflect user needs, avoiding ambiguity and inaccurate query results. The domain ontology in the dynamic knowledge base clarifies the terms, entity relationships, and hierarchical structure in the business scenario. Combined with the complex logic in the query elements, the system can automatically determine the logic and calculation method of data analysis by matching predefined indicator calculation rules and dimensional relationships, thereby ensuring that the query results meet actual business needs. Real-time metadata ensures the timeliness of query instructions, and the system can quickly synchronize the table structure, field meanings, and calculation rules of the database. When there are non-standard or ambiguous expressions in the query request, the term thesaurus built into the knowledge base can perform fuzzy matching and error correction, improving the system's fault tolerance and making the query process more intelligent and flexible, capable of handling diverse input forms. Through similarity matching, the system can transform the query element set into a computer-understandable structured query instruction.

[0079] In some embodiments of the present invention, for step S4, the query instruction is input into a preset statement optimization model, and the statement optimization model outputs the query statement;

[0080] The information in the query command may contain unnecessary repetitions or irrelevant parts, and redundancy needs to be removed by optimizing the model to ensure the simplicity of the query statement;

[0081] The query optimization model adjusts the query according to the intent of the query command to ensure that the output query accurately reflects the user's needs.

[0082] By adjusting the structure of the query statements to conform to the standard syntax of database query languages, the system's processing compatibility and efficiency can be improved.

[0083] The process of the query optimization model outputting the query statement includes:

[0084] The query optimization model analyzes the syntax of the query command to check whether it conforms to the basic grammatical rules of the language. This process relies on syntax analysis tools to identify the subject, predicate, object, modifiers, etc. in the command to ensure that the generated query statement has a reasonable structure.

[0085] The query optimization model performs semantic analysis on query commands to understand the implicit meaning in the query; the model uses natural language processing technology to identify key concepts, entities, relationships and their interrelationships in the query; for complex query commands, the model attempts to abstract the core intent of the query, remove redundant information and retain necessary contextual information;

[0086] The query optimization model combines the user's query history and contextual information to improve the accuracy of query optimization; if the user has previously queried a certain type of information, the model may infer the user's preferences based on this historical data, thereby optimizing the current query.

[0087] After completing the syntactic and semantic analysis, the model generates query statements with clearer structure and more explicit semantics through reconstruction, supplementation, or transformation.

[0088] Finally, the query optimization model will generate the final optimized query statement based on the above analysis and processing.

[0089] In this embodiment, the statement optimization model effectively removes unnecessary repetitions or irrelevant parts from query commands, ensuring the conciseness of the query statements and improving query efficiency. Through semantic analysis and user historical query data, the optimization model deeply understands the core intent of the query and adjusts it according to the context, ensuring that the output query statement more accurately reflects the user's needs and reducing the possibility of fuzzy queries. Through syntactic and semantic analysis, the optimization model automatically adjusts the structure of the query statement to conform to the standard syntax of database query languages, improving the system's processing compatibility and query efficiency, and ensuring that the system can execute query operations more smoothly. Based on user historical queries and contextual information optimization, the statement optimization model can more accurately meet user needs in different query scenarios, avoiding users repeatedly entering or modifying query commands, and improving the system's intelligence and convenience. Combined with natural language processing technology, the model can identify key concepts, entities, and their relationships in the query, and abstract complex queries through deep analysis, enabling the system to understand and respond to more complex query requirements, thereby improving the system's ability to handle complex queries.

[0090] In some embodiments of the present invention, for step S5, based on the query statement and the preset dynamic metadata management model, cross-system data automatic association query is performed to obtain the response result in response to the user's query request;

[0091] The dynamic metadata management model can synchronize metadata information of various business systems in real time; maintain data lineage, record the data flow path and conversion rules between different systems, and ensure logical consistency when querying across systems; support dynamic updates of metadata, and when the data structure of a business system changes, the model can complete the update within 5 minutes to ensure that the query statement matches the actual data storage logic;

[0092] Based on the system association information recorded by the dynamic metadata management model, the data sources involved in the query statement are analyzed, the query statement is broken down into sub-query tasks adapted to each system, and routed to the corresponding data interface.

[0093] For data that requires cross-system joint queries, the model automatically processes and concatenates the data according to predefined association rules;

[0094] Call the real-time data interfaces of each system to execute subqueries and obtain the raw data results;

[0095] The raw data returned by multiple systems is processed to unify the format, check for conflicts, and remove redundant data, generating a structured integrated result.

[0096] The integrated structured data is transformed into user-understandable responses, including structured data tables, visualizations, and natural language analysis results.

[0097] In this embodiment, the dynamic metadata management model can synchronize the metadata of various business systems in real time, ensuring logical consistency of data queries across different systems, effectively maintaining data lineage, and ensuring the accuracy and consistency of results when querying across systems. When the data structure of a business system changes, the model can update within 5 minutes, ensuring that the query statement matches the actual data storage logic, thereby reducing inconsistencies and potential errors caused by structural changes. By analyzing the data sources involved in the query statement, the model automatically decomposes the query statement into sub-query tasks adapted to each system, and routes the tasks to the corresponding data interfaces according to the system association information, improving query efficiency. The model can automatically process the data splicing of cross-system joint queries according to predefined association rules, ensuring that no errors or omissions occur during the data integration process. The model performs format unification, conflict verification, and redundant data removal on the original data returned by multiple systems, ensuring that the integrated data is accurate and standardized, and improving the reliability of the final query results. The integrated structured data can not only be converted into tables and visualization charts, but also generate natural language analysis conclusions, ensuring that end users can understand the query results more intuitively and easily.

[0098] like Figure 2 As shown, the present invention also provides a data result acquisition system based on large model analysis, which specifically includes the following modules;

[0099] The query request retrieval module retrieves the query request input by the user.

[0100] The semantic parsing module performs semantic parsing on the query request to obtain the query element set;

[0101] The similarity matching module substitutes the query element set into the preset dynamic knowledge base, performs similarity matching between the query element set and the dynamic knowledge base, and obtains the query command.

[0102] The query statement optimization module takes the query command as input into the preset statement optimization model, and the statement optimization model outputs the query statement.

[0103] The cross-system data query module automatically correlates and queries data across systems based on the query statement and the preset dynamic metadata management model, and obtains the response results in response to the user's query request.

[0104] In this embodiment, it breaks through the limitations of traditional fixed keyword matching or simple rule engines, and can deeply understand the implicit semantics and complex logic in query requests; it supports automatic conversion between business terms and colloquial expressions, and has a high accuracy rate in recognizing industry terms and proper nouns, avoiding errors in query element extraction due to misunderstandings of terminology; it supports multimodal input, including text, voice, and images, to meet users' query needs in different scenarios; it can handle complex queries such as multi-dimensional combined queries and trend predictions, no longer limited to simple single-table queries; at the same time, with the help of a dynamic knowledge base and a cross-system data query module, it realizes automatic cross-system data association queries without the need for manual configuration of association relationships, effectively solving the data silo problem; the query statement optimization module, based on a preset statement optimization model, can automatically optimize the execution of query statements. Efficiency is improved, reducing query time. Furthermore, combined with a query result caching mechanism, repeated accesses to the same question can directly return historical results, further shortening response time from hours or even days to seconds. The cross-system data query module automatically performs correlated queries based on a preset dynamic metadata management model, eliminating complex manual processes and significantly improving the efficiency of data query and integration. It lowers the operational threshold for users, reduces reliance on IT department data query support, and allows non-technical personnel to independently complete complex data analysis, reducing user training costs and enterprise labor costs. It features sensitive data anonymization and full lifecycle management of operation logs, recording all elements such as query conditions, result data, and access time, facilitating audit traceability and reducing the risk of sensitive data leakage.

[0105] In a specific implementation, as one example, the similarity matching module further includes an error correction unit;

[0106] When users input queries, spelling errors may occur during text input, speech-to-text conversion, or image-to-text conversion. Spelling errors can cause the query elements to mismatch with the entries in the dynamic knowledge base, affecting the accuracy of the matching. The error correction unit first corrects the words in the query request through a spell correction algorithm. The corrected query elements will be able to match the content in the knowledge base more accurately.

[0107] In query requests, users sometimes use inaccurate or uncommon words to express the same intent. This semantic error can affect the matching effect of the query. The error correction unit needs to correct the query request through semantic understanding and mapping technology.

[0108] In cases of polysemous words or unclear context, query requests may be ambiguous; the error correction unit eliminates such ambiguity through contextual analysis and contextual reasoning.

[0109] In this embodiment, the error correction unit automatically corrects spelling errors that users may make when inputting queries using a spelling correction algorithm. This avoids mismatches between query elements and dynamic knowledge base entries caused by inaccurate spelling, thus improving matching accuracy. Users sometimes use uncommon or inaccurate words to express the same intent. The error correction unit can correct these semantic errors through semantic understanding and mapping technology, enabling more accurate matching between query requests and knowledge base entries, thereby enhancing query performance. In cases of polysemous words or unclear context, the error correction unit effectively eliminates ambiguity in query requests through contextual analysis and contextual reasoning, ensuring that the system understands the user's true intent and improving matching accuracy.

[0110] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for obtaining data results based on large model analysis, characterized in that, include: Get the query request input by the user; Perform semantic parsing on the query request to obtain the query element set; Substitute the query element set into the preset dynamic knowledge base, perform similarity matching between the query element set and the dynamic knowledge base, and obtain the query instruction. Input the query command into the preset statement optimization model, and the statement optimization model outputs the query statement; Based on the query statement and the preset dynamic metadata management model, automatic cross-system data association queries are performed to obtain response results in response to user query requests; The query element set is matched with the dynamic knowledge base for similarity to obtain the query command: The query elements are mapped in a dynamic knowledge base to identify the standardized terms and business definitions corresponding to the elements. Logical matching is performed using predefined indicator calculation rules and dimension relationships in the knowledge base to determine the required data analysis logic and calculation methods; The built-in thesaurus of terms in the knowledge base is used to perform fuzzy matching and error correction on the parts of the query elements that have non-standard or ambiguous expressions. The query element set is converted into query instructions by using mapping results, logical matching results, and term thesaurus query results; Based on the query statement and the preset dynamic metadata management model, automatic cross-system data association queries are performed to obtain the response results in response to the user's query request: Based on the system association information recorded by the dynamic metadata management model, the data sources involved in the query statement are analyzed, the query statement is broken down into sub-query tasks adapted to each system, and routed to the corresponding data interface. For data that requires cross-system joint queries, data concatenation is processed according to predefined association rules; Call the real-time data interfaces of each system to execute subqueries and obtain the raw data results; The raw data returned by multiple systems is standardized to generate a structured integrated result; Transform the integrated results into responses that users can understand.

2. The method for obtaining data results based on large model analysis according to claim 1, characterized in that, The query request includes at least one of text requests, voice requests, and image requests.

3. The method for obtaining data results based on large model analysis according to claim 1, characterized in that, The query element set includes query intent, key entities, logical relationships, contextual information, and business terms.

4. The method for obtaining data results based on large model analysis according to claim 1, characterized in that, The query command includes the data query scope, calculation logic, and related dimensions.

5. The method for obtaining data results based on large model analysis according to claim 1, characterized in that, The process of the query optimization model outputting the query statement includes: The query optimization model analyzes the syntax of the query command and examines its basic syntax rules. Perform semantic analysis on query commands to obtain their implicit meanings; The query commands are processed according to user preferences by combining the user's query history and context information; The query optimization model generates the final optimized query statement based on syntactic and semantic analysis and user preference processing results.

6. The method for obtaining data results based on large model analysis according to claim 5, characterized in that, The response results in response to user query requests consist of structured data tables, visual charts, and natural language analysis conclusions.

7. A data result acquisition system based on large model analysis, used to execute the method of claim 1, characterized in that, include: The query request retrieval module retrieves the query request input by the user. The semantic parsing module performs semantic parsing on the query request to obtain the query element set; The similarity matching module substitutes the query element set into the preset dynamic knowledge base, performs similarity matching between the query element set and the dynamic knowledge base, and obtains the query command. The query statement optimization module takes the query command as input into the preset statement optimization model, and the statement optimization model outputs the query statement. The cross-system data query module automatically correlates and queries data across systems based on the query statement and the preset dynamic metadata management model, and obtains the response results in response to the user's query request.

8. The data result acquisition system based on large model analysis according to claim 7, characterized in that, The similarity matching module also includes an error correction unit.

Images