An artificial intelligence-based literature information automatic evolution construction method and system

Abstract

This invention relates to an artificial intelligence-based method and system for the automatic evolution and construction of document information, comprising: receiving documents; intelligently processing the documents to obtain structured information; and forming an automatically evolving knowledge base based on the structured information. This invention can improve the efficiency and accuracy of document information processing and knowledge base construction, achieve dynamic updates of the knowledge base, adapt to multiple scenario requirements, and balance data security and processing quality.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to an automatic evolution and construction method and system for document information based on artificial intelligence. Background Technology

[0002] In today's information explosion era, the number of text documents such as scientific research literature and industry reports is growing exponentially. Traditional methods of document information processing and knowledge base construction mainly rely on manual sorting, classification and input, which has problems such as low efficiency, high labor costs, inaccurate information extraction and lagging knowledge base updates.

[0003] In existing technologies, some document processing methods use simple keyword matching and text segmentation for information extraction, but they cannot achieve a deep understanding of the semantics of the documents, making it difficult to extract implicit and related information, and they cannot achieve automatic updates and evolution of the knowledge base. Meanwhile, existing AI-based document processing methods often suffer from drawbacks such as single large-model calls, cumbersome information processing workflows, and loose knowledge base structures, failing to meet users' needs for efficient utilization and dynamic updates of document information. For example, knowledge bases generated by traditional methods often suffer from inconsistent entity descriptions and redundant relationships, significantly reducing their practicality; while some intelligent processing methods can extract basic information, they lack effective information aggregation and standardization mechanisms, failing to form a structured, reusable, high-quality knowledge base. Furthermore, existing technologies lack flexible multi-model adaptation capabilities, making it difficult to select appropriate processing models based on document type, data sensitivity, and other requirements. This either compromises the security of core document data or makes it difficult to balance processing efficiency and quality.

[0004] Therefore, there is an urgent need for a method that can automatically receive and intelligently process document information, and automatically evolve and construct a structured knowledge base, in order to solve the technical pain points of existing technologies such as low efficiency, poor accuracy, untimely knowledge base updates, and weak adaptability. Summary of the Invention

[0005] This invention provides an automatic evolution and construction method and system for literature information based on artificial intelligence, in order to overcome the shortcomings of existing technologies.

[0006] This invention provides an artificial intelligence-based method for automatically evolving and constructing documentary information, comprising: S1: Receive documents; wherein the documents are text documents filtered according to preset format requirements; S2: Perform intelligent processing on the document to obtain structured information; S3: Based on the structured information, form a knowledge base that can evolve automatically.

[0007] According to the present invention, an automatic evolution and construction method for document information based on artificial intelligence is provided. In step S1, the document is a text-based document selected according to preset format requirements. The document format includes at least one of PDF, DOCX, TXT, and MD.

[0008] According to the present invention, an automatic evolutionary construction method for literature information based on artificial intelligence, step S1 further includes: S11: Receive documents through the document receiving module. The receiving method includes at least one of the following: local file upload, web link crawling, and API call import. S12: Preprocess the received documents, including format parsing, redundant information removal, and encoding unification, converting the documents into a unified text format that can be recognized by the large model. S13: Record the basic metadata of the document, which includes at least one of the following: document title, author, publication date, source, and format type; S14: Select a suitable large model based on the sensitivity of the literature data and processing requirements. The large model includes locally deployed open-source models and cloud-based closed-source models.

[0009] According to the present invention, an automatic evolution and construction method for literature information based on artificial intelligence is provided, wherein the open-source model includes Llama3-70B, and the closed-source model includes at least one of GPT-4-turbo, Tongyi Qianwen Max, DeepSeek-V2, and Baichuan5; the redundant information includes at least one of blank pages, watermarks, and irrelevant advertisements.

[0010] According to the present invention, an automatic evolutionary construction method for literature information based on artificial intelligence is provided, wherein the intelligent processing in step S2 includes at least one of the following: Deep text understanding: Through natural language processing technology, semantic analysis and contextual analysis are performed on the document text to understand at least one of the core theme, research content, technical solutions, and conclusions of the document, and to identify technical terms and ambiguous sentences in the document; Key information extraction: Automatically extract at least one of the following from the literature: keywords, core viewpoints, technical parameters, experimental data, conclusions, references, entities and relationships between entities, and perform preliminary normalization on the extracted entities; Information classification and filtering: Based on the content attributes, subject areas, and information types of the documents, the extracted key information is automatically classified, and invalid and duplicate information is filtered out; Information verification and correction: Combining the knowledge system trained by the large model, identify errors and contradictions in the information, make preliminary corrections based on the context of the documents, and mark ambiguous information that cannot be determined as awaiting manual verification. Dynamic adaptation and optimization: Based on the document type, the processing strategy is automatically adjusted to extract the core information of the corresponding document type.

[0011] According to the present invention, an automatic evolution and construction method for literature information based on artificial intelligence is provided, wherein the literature type includes at least one of scientific research literature, patent literature, and industry reports; and the entity includes at least one of people, institutions, technical terms, and product names.

[0012] According to the present invention, an automatic evolutionary construction method for literature information based on artificial intelligence, step S3 further includes: S31: Use at least one of knowledge graphs and relational databases to store structured information in a structured manner, where entities serve as nodes in the knowledge base and the relationships between entities serve as connections between nodes; S32: Through the knowledge base update module, newly received and processed document information is automatically integrated into the existing knowledge base, realizing the automatic evolution of the knowledge base; S33: Provides knowledge base retrieval and interaction functions, wherein the retrieval methods include at least one of keyword retrieval, semantic retrieval, and association retrieval, and supports information export function.

[0013] According to the present invention, an automatic evolution construction method for literature information based on artificial intelligence is provided. The automatic evolution in step S32 includes adding new entities, supplementing entity relationships, updating existing information, and periodically optimizing the knowledge base; the periodic optimization includes removing outdated information and correcting erroneous information.

[0014] According to the present invention, an automatic evolution and construction method for document information based on artificial intelligence is provided. In step S31, entities extracted from multiple documents are normalized through semantic clustering, similarity retrieval and deduplication, and the normalized entity name is selected as the standard representative.

[0015] According to the artificial intelligence-based automatic evolution and construction method for literature information provided by the present invention, step S3 further includes: The information sources and feature associations of the knowledge base are interpreted in a personalized manner using a global interpretation method, clarifying the contribution of each structured information to the knowledge base.

[0016] This invention also provides an artificial intelligence-based automatic evolution and construction system for literature information, comprising: The receiving module is used to receive documents; wherein the documents are text documents filtered according to preset format requirements. The processing module is used to intelligently process the documents to obtain structured information; A building module is used to form an automatically evolving knowledge base based on the structured information.

[0017] This invention provides an AI-based method and system for the automatic evolution and construction of document information. It aims to develop an interpretable and highly efficient intelligent document processing and knowledge base construction scheme, combining the deep semantic understanding capabilities of large models with structured knowledge base construction to achieve automatic reception, intelligent processing, and dynamic evolution of document information and the knowledge base. Furthermore, this invention employs a flexible multi-model adaptation mechanism to balance data security and processing quality, and utilizes a global interpretation method to clarify the contribution of each structured piece of information, thus verifying the interpretability of the scheme.

[0018] This invention, building upon traditional document processing methods and leveraging the advantages of large-scale artificial intelligence models, achieves automated and intelligent processing of document information, addressing the pain points of low efficiency and high labor costs associated with traditional methods. Recognizing the limitations of traditional intelligent processing methods in terms of their singularity and information extraction capabilities, this invention employs a multi-dimensional intelligent processing workflow, including deep text understanding, key information extraction, classification and filtering, verification and correction, and dynamic adaptation and optimization, combined with entity normalization processing to improve the accuracy of information extraction. Simultaneously, through an automatically evolving knowledge base construction method, it achieves automatic integration of new document information and regular optimization of the knowledge base, combining knowledge graphs and relational databases to construct a structured and highly available knowledge base. Furthermore, this invention utilizes a flexible multi-model adaptation mechanism, selecting the appropriate large-scale model based on the sensitivity of the document data and processing needs, balancing data security and processing quality, and adapting to the document information management needs of various scenarios such as research, enterprises, and government. Through a global interpretation method, it clarifies the contribution of each structured piece of information to the knowledge base, more accurately achieving efficient utilization of document information and dynamic updates to the knowledge base, making it suitable for widespread application. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this 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 some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 A schematic diagram of a method for automatically evolving and constructing document information based on artificial intelligence, provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of an artificial intelligence-based automatic evolution and construction system for document information, provided as an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, embodiments of this invention, and should not be construed as limiting the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention. In the description of this invention, it should be understood that the terminology used is for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0022] The embodiments of the present invention are described below with reference to the figures.

[0023] like Figure 1 As shown, this invention provides an automatic evolutionary construction method for literature information based on artificial intelligence, comprising: S1: Receive documents; wherein the documents are text documents filtered according to preset format requirements.

[0024] In step S1, the main function is to receive various text-based documents that meet the format requirements through the built document receiving module. It supports the automatic reception of documents in multiple formats, including but not limited to common text formats such as PDF, DOCX, TXT, and MD. The receiving methods include local file upload, web link crawling, and API call import, which can meet the document acquisition needs in different scenarios.

[0025] In step S1, the documents are text-based documents that have been filtered according to preset format requirements. Documents with standardized format and complete content are selected first, while documents that cannot be parsed or have invalid content are removed to ensure the smooth progress of subsequent intelligent processing.

[0026] Step S1 further includes: S11: Receive documents through the document receiving module. The receiving method includes at least one of the following: local file upload, web link crawling, and API call import. In a specific embodiment, in a scientific research scenario, research papers can be imported through an academic database interface; in an enterprise scenario, internal technical reports can be imported through local upload.

[0027] S12: Preprocess the received documents, including format parsing, redundant information removal, and encoding unification, converting the documents into a unified text format that can be recognized by the large model. Redundant information includes at least one of blank pages, watermarks, and irrelevant advertisements, and is uniformly encoded in UTF-8 to ensure that the large model can accurately identify and process the document content.

[0028] S13: Record the basic metadata of the document, which includes at least one of the following: document title, author, publication date, source, and format type; In one specific embodiment, the recording of metadata facilitates subsequent retrieval, classification, and management of the knowledge base, enabling the traceability of document information.

[0029] S14: Select a suitable large model based on the sensitivity of the literature data and processing requirements. The large model includes open-source models deployed locally and closed-source models in the cloud. Among them, the open-source model includes Llama3-70B, and the closed-source model includes at least one of GPT-4-turbo, Tongyi Qianwen Max, DeepSeek-V2, and Baichuan5; In a specific implementation, for data-sensitive documents (such as internal enterprise technical reports), the locally deployed Llama3-70B open-source model is selected to ensure that core data is not leaked; for high-quality document processing needs (such as scientific research papers), cloud-based closed-source models such as GPT-4-turbo and Tongyi Qianwen Max are selected to improve processing quality and efficiency; for scenarios with mainly Chinese documents, models with strong Chinese understanding capabilities such as DeepSeek-V2 and Baichuan5 are selected to optimize semantic understanding effects.

[0030] In one specific embodiment, this embodiment collects 100 research papers in the field of computer science from research institutions that need to build a scientific literature knowledge base. These papers are in PDF format (80 papers), DOCX format (15 papers), and TXT format (5 papers). 60 papers were imported through an academic database interface, and 40 papers were uploaded locally. The 100 received papers are preprocessed: 3 papers containing watermarks and 2 papers with excessive blank pages are removed. The remaining 95 papers are converted to UTF-8 encoded TXT format, and metadata such as title, author, journal, publication date, and source are recorded for each paper. Since this batch of papers involves non-sensitive scientific data, the GPT-4-turbo model is selected as the core processing model to ensure processing quality and efficiency.

[0031] S2: Perform intelligent processing on the document to obtain structured information.

[0032] After receiving and preprocessing the documents in step S1, step S2 mainly involves inputting the preprocessed document text into the selected large-scale artificial intelligence model. The large-scale model performs multi-dimensional intelligent processing on the documents, converting the unstructured document text into structured information, thus providing a foundation for the subsequent construction of the knowledge base.

[0033] The intelligent processing in step S2 includes at least one of the following: Deep text understanding: Through natural language processing technology, semantic analysis and contextual analysis are performed on the document text to understand at least one of the core themes, research content, technical solutions, and conclusions of the document, and to identify technical terms and ambiguous sentences in the document.

[0034] In one specific embodiment, when processing scientific research papers in the field of computer science, it is necessary to accurately understand the research direction (such as artificial intelligence, machine learning), core research questions, research methods, experimental design, core conclusions, etc., and identify professional terms in the paper (such as Transformer, CNN, dataset names, etc.) to avoid semantic misunderstandings.

[0035] Key information extraction: Automatically extract at least one of the following from the literature: keywords, core viewpoints, technical parameters, experimental data, conclusions, references, entities and relationships between entities, and perform preliminary normalization on the extracted entities; where entities include at least one of people, organizations, technical terms, and product names.

[0036] In one specific embodiment, keywords (such as "Transformer model" and "image classification"), experimental data (such as "accuracy reached 98.2% on the ImageNet dataset"), entity information (such as "Transformer model", "ImageNet dataset", and "a research team") and relationships between entities (such as "a research team proposed an improved Transformer model") from research papers are extracted, and "Transformer" and "Transformer model" are unified as "Transformer model" to avoid different expressions of the same entity being extracted repeatedly.

[0037] Information classification and filtering: Based on the content attributes, subject areas, and information types of the documents, the extracted key information is automatically classified, and invalid and duplicate information is filtered out.

[0038] In one specific embodiment, the information in the scientific literature is divided into five categories: “research background”, “research methods”, “experimental data”, “core conclusions” and “references”. Duplicate references and invalid experimental data (such as outliers) are removed, and the core and valid information is retained.

[0039] Information verification and correction: Combining the knowledge system trained by the large model, errors and contradictions in the information are identified, and preliminary corrections are made based on the context of the documents. Ambiguous information that cannot be determined is marked as awaiting manual verification.

[0040] In one specific implementation, a contradiction was found between "accuracy rate 98.2%" and the context "accuracy rate 97.8%" in a certain paper, and it was marked as requiring manual verification; the extracted professional terms were standardized to ensure the accuracy of the information.

[0041] Dynamic adaptation and optimization: Based on the document type, the processing strategy is automatically adjusted to focus on extracting the core information of the corresponding document type; the document type includes at least one of scientific research documents, patent documents, and industry reports.

[0042] In a specific embodiment, when processing patent documents, the focus is on extracting the technical features from the claims and specifications; when processing academic papers, the focus is on extracting the research methods, experimental data, and conclusions; and when processing industry reports, the focus is on extracting industry trends, data statistics, and core viewpoints.

[0043] In a specific embodiment provided by this invention, the preprocessed texts of 95 scientific research papers were batch-input into the GPT-4-turbo large model. After the large model completed intelligent processing, a total of 285 keywords, 190 core viewpoints, 120 sets of experimental data, 320 entity information, and 250 sets of relationships between entities were extracted. 35 invalid information entries and 42 duplicate information entries were filtered out, and 18 information entries were marked for manual verification. Finally, 810 pieces of structured information were obtained, providing high-quality basic data for the subsequent construction of the knowledge base.

[0044] S3: Based on the structured information, form a knowledge base that can evolve automatically.

[0045] In step S3, based on the structured information obtained in step S2, a knowledge base is constructed using a structured storage method. The knowledge base is automatically evolved through a knowledge base update module, while providing retrieval and interaction functions to improve the practicality and ease of use of the knowledge base.

[0046] Step S3 further includes: S31: Use at least one of knowledge graphs and relational databases to store structured information in a structured manner, where entities serve as nodes in the knowledge base and the relationships between entities serve as connections between nodes.

[0047] Entities extracted from multiple documents are standardized through semantic clustering, similarity retrieval, and deduplication, and standardized entity names are selected as standard representatives.

[0048] In one specific embodiment, a knowledge graph is used, with extracted entities (such as "Transformer model" and "ImageNet dataset") as nodes and relationships between entities (such as "applied to" and "proposed") as connections between nodes, to construct a hierarchical knowledge graph. At the same time, a relational database is used to store document metadata and classified key information to achieve structured storage. Similar entities such as "improved Transformer" and "optimized Transformer" are merged, and "improved Transformer model" is selected as the standard name to improve the standardization of the knowledge base.

[0049] S32: Through the knowledge base update module, newly received and processed literature information is automatically integrated into the existing knowledge base, realizing the automatic evolution of the knowledge base.

[0050] The automatic evolution includes adding new entities, supplementing entity relationships, updating existing information, and periodically optimizing the knowledge base; the periodic optimization includes removing outdated information and correcting erroneous information.

[0051] In one specific embodiment, when a user subsequently uploads new research papers in the field of computer science, the document receiving module automatically receives and preprocesses them. After the large model completes intelligent processing, it automatically integrates the newly extracted entities, relationships, experimental data, and other information into the existing knowledge base, adds nodes such as "new model" and "new dataset," and supplements the relationships. The knowledge base is reviewed monthly through the large model to remove outdated experimental data and correct errors, ensuring the timeliness of the knowledge base.

[0052] S33: Provides knowledge base retrieval and interaction functions, wherein the retrieval methods include at least one of keyword retrieval, semantic retrieval, and association retrieval, and supports information export function.

[0053] In one specific embodiment, users can quickly obtain the core information and entity relationships of relevant papers through keyword search (such as "Transformer model") and semantic search (such as "deep learning model for image classification"); the search results can be exported to PDF and Excel formats to meet the needs of researchers for literature organization and analysis.

[0054] Step S3 is followed by: The information sources and feature associations of the knowledge base are interpreted in a personalized manner using a global interpretation method, clarifying the contribution of each structured information to the knowledge base.

[0055] In one specific embodiment, the contribution of each structured piece of information in the knowledge base is evaluated using Shapley Additive Interpretation (SHAP) values ​​to generate a ranking of feature importance, thereby clarifying which information contributes significantly to the usability and accuracy of the knowledge base and improving the interpretability of the solution.

[0056] In a specific embodiment of this invention, a knowledge base for scientific research papers in the field of computer science is constructed based on 810 pieces of structured information, using a combination of knowledge graph and relational database. The knowledge graph contains 320 entities and 250 relationships between entities, while the relational database stores 95 pieces of document metadata and 685 categorized key information. A knowledge base update module automatically incorporates new document information, and the knowledge base is optimized monthly. It provides three retrieval methods: keyword retrieval, semantic retrieval, and association retrieval, and supports exporting search results to PDF and Excel formats. SHAP values ​​are used for global interpretation to clarify the contribution of each piece of structured information to the knowledge base, improving its interpretability.

[0057] like Figure 2 As shown, the present invention also provides an artificial intelligence-based automatic evolution and construction system for literature information, comprising: The receiving module 100 is used to receive documents; wherein the documents are text documents filtered according to preset format requirements. Processing module 200 is used to intelligently process the document to obtain structured information; Construction module 300: Used to form an automatically evolving knowledge base based on the structured information.

[0058] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0059] This invention provides an AI-based method and system for the automatic evolution and construction of literature information. Through large-scale intelligent model processing and structured knowledge base construction, it achieves automated and intelligent processing of literature information and dynamic evolution of the knowledge base. It can perform individualized and accurate processing of literature information. Based on the original literature processing workflow, it does not require additional labor costs and equipment investment, making it easy to promote and apply widely. Compared with the prior art, this invention has higher processing efficiency and greater accuracy, and is more conducive to users' efficient use of literature information and the dynamic accumulation of knowledge.

[0060] This invention provides an AI-based method and system for the automatic evolution and construction of document information. By combining large-scale intelligent processing with structured knowledge base construction, and utilizing a multi-dimensional intelligent processing flow, it obtains high-quality structured information, thereby improving the accuracy and reliability of knowledge base construction. Secondly, before receiving documents, this invention ensures document quality through format screening and preprocessing, while flexible adaptation using multiple models balances data security and processing quality, contributing to improved adaptability and practicality. Thirdly, in the intelligent processing stage, through a multi-step processing and screening process, redundant and irrelevant information is effectively removed, retaining the structured information most relevant to the core content of the documents, thus improving the quality of the knowledge base. Then, in the knowledge base construction stage, a combination of knowledge graphs and relational databases is used to achieve structured storage, and automatic evolution is achieved through a knowledge base update module, ensuring the timeliness and completeness of the knowledge base. After the knowledge base is constructed, this invention uses a global interpretation method to provide personalized interpretation of the knowledge base, clarifying the contribution of each structured information, ensuring the interpretability of the solution, facilitating the verification of the solution's performance in practical applications, and allowing for adjustments and optimizations as needed.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for automatically evolving and constructing document information based on artificial intelligence, characterized in that, include: S1: Receive documents; wherein the documents are text documents filtered according to preset format requirements; S2: Perform intelligent processing on the document to obtain structured information; S3: Based on the structured information, form a knowledge base that can evolve automatically.

2. The method for automatically evolving and constructing document information based on artificial intelligence according to claim 1, characterized in that, Step S1 further includes: S11: Receive documents through the document receiving module. The receiving method includes at least one of the following: local file upload, web link crawling, and API call import. S12: Preprocess the received documents, including format parsing, redundant information removal, and encoding unification, converting the documents into a unified text format that can be recognized by the large model. S13: Record the basic metadata of the document, which includes at least one of the following: document title, author, publication date, source, and format type; S14: Select a suitable large model based on the sensitivity of the literature data and processing requirements. The large model includes locally deployed open-source models and cloud-based closed-source models.

3. The method for automatically evolving and constructing document information based on artificial intelligence according to claim 2, characterized in that, The open-source model includes Llama3-70B, and the closed-source model includes at least one of GPT-4-turbo, Tongyi Qianwen Max, DeepSeek-V2, and Baichuan5; the redundant information includes at least one of blank pages, watermarks, and irrelevant advertisements.

4. The method for automatically evolving and constructing document information based on artificial intelligence according to claim 1, characterized in that, The intelligent processing in step S2 includes at least one of the following: Deep text understanding: Through natural language processing technology, semantic analysis and contextual analysis are performed on the document text to understand at least one of the core theme, research content, technical solutions, and conclusions of the document, and to identify technical terms and ambiguous sentences in the document; Key information extraction: Automatically extract at least one of the following from the literature: keywords, core viewpoints, technical parameters, experimental data, conclusions, references, entities and relationships between entities, and perform preliminary normalization on the extracted entities; Information classification and filtering: Based on the content attributes, subject areas, and information types of the documents, the extracted key information is automatically classified, and invalid and duplicate information is filtered out; Information verification and correction: Combining the knowledge system trained by the large model, identify errors and contradictions in the information, make preliminary corrections based on the context of the documents, and mark ambiguous information that cannot be determined as awaiting manual verification. Dynamic adaptation and optimization: Based on the document type, the processing strategy is automatically adjusted to extract the core information of the corresponding document type.

5. The method for automatically evolving and constructing document information based on artificial intelligence according to claim 4, characterized in that, The document types include at least one of scientific research documents, patent documents, and industry reports; the entities include at least one of people, organizations, technical terms, and product names.

6. The method for automatically evolving and constructing document information based on artificial intelligence according to claim 1, characterized in that, Step S3 further includes: S31: Use at least one of knowledge graphs and relational databases to store structured information in a structured manner, where entities serve as nodes in the knowledge base and the relationships between entities serve as connections between nodes; S32: Through the knowledge base update module, newly received and processed document information is automatically integrated into the existing knowledge base, realizing the automatic evolution of the knowledge base; S33: Provides knowledge base retrieval and interaction functions, wherein the retrieval methods include at least one of keyword retrieval, semantic retrieval, and association retrieval, and supports information export function.

7. The method for automatically evolving and constructing document information based on artificial intelligence according to claim 6, characterized in that, The automatic evolution in step S32 includes adding new entities, supplementing entity relationships, updating existing information, and periodically optimizing the knowledge base; the periodic optimization includes removing outdated information and correcting erroneous information.

8. The method for automatically evolving and constructing document information based on artificial intelligence according to claim 6, characterized in that, In step S31, entities extracted from multiple documents are standardized through semantic clustering, similarity retrieval, and deduplication, and standardized entity names are selected as standard representatives.

9. The method for automatically evolving and constructing document information based on artificial intelligence according to claim 1, characterized in that, Step S3 is followed by: The information sources and feature associations of the knowledge base are interpreted in a personalized manner using a global interpretation method, clarifying the contribution of each structured information to the knowledge base.

10. An artificial intelligence-based automatic evolution and construction system for document information, characterized in that, include: The receiving module is used to receive documents; The documents mentioned therein are text-based documents that have been filtered according to preset format requirements; The processing module is used to intelligently process the documents to obtain structured information; A building module is used to form an automatically evolving knowledge base based on the structured information.

Images