Document analysis system, method and equipment based on knowledge graph and medium

Abstract

The invention discloses a literature analysis system, method, equipment and medium based on a knowledge graph, the system integrates a literature knowledge graph construction module, an entity information extraction module and a query response module, the literature knowledge graph construction module is used for extracting entities, relationships and attributes from literatures, and the entity information extraction module is used for extracting the entities, the relationships and the attributes from the literatures; integrating the entities, the relationships and the attributes to construct a literature knowledge graph; the entity information extraction module is used for performing entity identification on input query information to obtain a query entity related to the query information, and retrieving data information related to the query entity from the literature knowledge graph based on the query entity; and the query response module is used for generating a structured query list according to the data information and generating a query result according to the structured query list and the literature knowledge graph. According to the embodiment of the invention, the effect and efficiency of literature analysis are remarkably improved, so that a more comprehensive and deep literature analysis result can be provided for a user.

Description

Technical Field

[0001] The embodiments of the present application relate to but are not limited to the field of information retrieval technology, and in particular to a document analysis system, method, device and medium based on knowledge graph. Background Art

[0002] Most current literature systems are limited to processing a single data perspective. For example, they are usually visualized based on a certain dimension such as citations, authors, or research topics. This single-perspective analysis method is difficult to fully cope with the increasingly complex and diverse literature data. Therefore, these systems often fail to provide users with a comprehensive understanding of literature data. Although some innovative interactive visualization tools have begun to allow users to dynamically explore data sets from multiple perspectives, existing analysis methods generally lack the ability to automatically integrate data from different perspectives and conduct in-depth analysis. In addition, as the complexity and heterogeneity of literature data increase, traditional visualization and analysis methods are no longer sufficient to meet the current needs of scientific literature data analysis.

[0003] Therefore, there is an urgent need for more advanced analytical methods that can automatically integrate information from different perspectives in order to more comprehensively reveal the complex relationships in the literature. Summary of the invention

[0004] The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.

[0005] The embodiment of the present application provides a document analysis system based on a knowledge graph, which significantly improves the effect and efficiency of document analysis, thereby providing users with more comprehensive and in-depth document analysis results.

[0006] In the first aspect, an embodiment of the present application provides a document analysis system based on a knowledge graph, including: a document knowledge graph construction module, used to extract entities, relationships and attributes from documents, and integrate the entities, the relationships and the attributes to construct a document knowledge graph; an entity information extraction module, used to perform entity recognition on input query information, obtain query entities related to the query information, and retrieve data information related to the query entity from the document knowledge graph based on the query entity; a query response module, used to generate a structured query list based on the data information, and generate query results based on the structured query list and the document knowledge graph.

[0007] In combination with the first aspect, in one embodiment of the present application, the document knowledge graph construction module includes: a data preprocessing unit, which is used to collect document data from databases, document libraries and network resources, and perform cleaning and standardization processing to obtain standard document data; a multi-dimensional processing unit, which is used to parse the text information in the standard document data, obtain entities, relationships and attributes, and construct a graph structure based on the entities, the relationships and the attributes; a graph generation unit, which is used to extract feature information of the graph structure and construct a document knowledge graph based on the feature information.

[0008] In combination with the first aspect, in one embodiment of the present application, the entity information extraction module includes: a retrieval enhancement unit, which is used to perform feature matching based on the feature information of the query entity and the graph structure to obtain a matching result, and based on the matching result, query the document data to obtain data information related to the query entity.

[0009] In combination with the first aspect, in one embodiment of the present application, the system also includes a data display module, which includes: a multi-perspective visualization unit, used to generate a variety of visualization charts based on the query results and display the visualization charts, the multiple visualization charts including at least one of a keyword cloud, a department pie chart, a year stacked bar chart, and an author tree chart.

[0010] In combination with the first aspect, in an embodiment of the present application, the system further includes: an interactive exploration module, configured to dynamically query the visualization chart according to a plurality of interactive operations inputted.

[0011] In the second aspect, an embodiment of the present application provides a document analysis method based on a knowledge graph, including: extracting entities, relationships and attributes from documents, integrating the entities, the relationships and the attributes to construct a document knowledge graph; performing entity recognition on the input query information to obtain query entities related to the query information; retrieving data information related to the query entity from the document knowledge graph based on the query entity; generating a structured query list based on the data information, and generating query results based on the structured query list and the document knowledge graph.

[0012] In combination with the second aspect, in one embodiment of the present application, the entities, relationships and attributes are extracted from the documents, and the entities, relationships and attributes are integrated to construct a document knowledge graph, including: collecting document data from multiple sources, and performing cleaning and standardization processing to obtain standard document data, wherein the multiple sources include databases, document libraries and network resources; parsing the text information in the standard document data to obtain entities, relationships and attributes, and constructing a graph structure based on the entities, relationships and attributes; extracting feature information of the graph structure, and constructing a document knowledge graph based on the feature information.

[0013] In combination with the second aspect, in one embodiment of the present application, the retrieving data information related to the query entity from the literature knowledge graph based on the query entity includes: performing feature matching based on feature information of the query entity and the graph structure to obtain a matching result; and querying the literature data based on the matching result to obtain data information related to the query entity.

[0014] On the other hand, an embodiment of the present application provides an electronic device, comprising: at least one processor; at least one memory for storing at least one program; when at least one of the programs is executed by at least one of the processors, the document analysis method as described above is implemented.

[0015] On the other hand, a computer-readable storage medium stores a computer program executable by a processor, and the computer program executable by the processor is used to implement the document analysis method as described above when executed by the processor.

[0016] The document analysis system based on the knowledge graph provided by the embodiment of the present application integrates a document knowledge graph construction module, an entity information extraction module and a query response module, wherein the document knowledge graph construction module is used to extract entities, relationships and attributes from documents, and integrate entities, relationships and attributes to construct a document knowledge graph. This module breaks the limitation of the traditional document system that only processes a single data perspective, so that the system can understand the document data more comprehensively and reveal the complex relationships and potential trends therein. The entity information extraction module is used to perform entity recognition on the input query information, obtain query entities related to the query information, and retrieve data information related to the query entity from the document knowledge graph based on the query entity. This process not only improves the analysis efficiency, but also reduces manual intervention, making the system more intelligent. The query response module is used to generate a structured query list based on the data information, and generate query results based on the structured query list and the document knowledge graph. The knowledge graph-based document analysis system of the embodiment of the present application significantly improves the effectiveness and efficiency of document analysis by comprehensively integrating document data, realizing automated and intelligent analysis, multi-angle integration and analysis, coping with complexity and heterogeneity, and providing user-friendly and interactive query results, thereby providing users with more comprehensive and in-depth document analysis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the structure of a knowledge graph-based document analysis system provided in an embodiment of the present application;

[0018] Figure 2 It is a workflow diagram of the document analysis system provided in the embodiment of the present application;

[0019] Figure 3 It is a flowchart of a literature knowledge graph construction module provided by a specific example of this application;

[0020] Figure 4 is a schematic diagram of a visualization interface of a document analysis system provided in an embodiment of the present application;

[0021] Figure 5 It is a flow chart of a document analysis method based on knowledge graph provided in an embodiment of the present application;

[0022] Figure 6 It is a schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0024] It should be noted that, although the logical order is shown in the flowchart, in some cases, the steps shown or described can be performed in a different order from that in the flowchart. The terms "first", "second", etc. in the specification, claims and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification for people familiar with this technology to understand and read, and are not used to limit the limiting conditions that can be implemented in this application, so they have no technical substantive significance. Any modification of the structure, change of the proportional relationship or adjustment of the size, without affecting the effects that can be produced by this application and the purposes that can be achieved, should still fall within the scope of the technical content disclosed in this application. At the same time, the terms such as "upper", "lower", "left", "right", "middle" and "one" quoted in this specification are only for the convenience of narration, and are not used to limit the scope of implementation of this application. The change or adjustment of the relative relationship should also be regarded as the scope of implementation of this application without substantial change of the technical content.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0026] Most current literature systems are limited to processing a single data perspective. For example, they are usually visualized based on a certain dimension such as citations, authors, or research topics. This single-perspective analysis method is difficult to fully cope with the increasingly complex and diverse literature data. Therefore, these systems often fail to provide users with a comprehensive understanding of literature data. Although some innovative interactive visualization tools have begun to allow users to dynamically explore data sets from multiple perspectives, existing analysis methods generally lack the ability to automatically integrate data from different perspectives and conduct in-depth analysis. In addition, as the complexity and heterogeneity of literature data increase, traditional visualization and analysis methods are no longer sufficient to meet the current needs of scientific literature data analysis.

[0027] Therefore, there is an urgent need for more advanced analytical methods that can automatically integrate information from different perspectives in order to more comprehensively reveal the complex relationships in the literature.

[0028] In view of this, the embodiment of the present application provides a document analysis system, document analysis method, electronic device and computer-readable storage medium based on knowledge graph, wherein the document analysis system integrates a document knowledge graph construction module, an entity information extraction module and a query response module, wherein the document knowledge graph construction module is used to extract entities, relationships and attributes from documents, and integrate entities, relationships and attributes to construct a document knowledge graph. This module breaks the limitation of traditional document systems that only process a single data perspective, so that the system can understand document data more comprehensively and reveal the complex relationships and potential trends therein. The entity information extraction module is used to perform entity recognition on the input query information, obtain query entities related to the query information, and retrieve data information related to the query entity from the document knowledge graph based on the query entity. This process not only improves the analysis efficiency, but also reduces manual intervention, making the system more intelligent. The query response module is used to generate a structured query list based on the data information, and generate query results based on the structured query list and the document knowledge graph. Compared with existing analysis methods, the knowledge graph-based document analysis system of the embodiment of the present application significantly improves the effectiveness and efficiency of document analysis by comprehensively integrating document data, realizing automated and intelligent analysis, multi-angle integration and analysis, coping with complexity and heterogeneity, and providing user-friendly and interactive query results, providing users with more comprehensive and in-depth document analysis results.

[0029] The embodiments of the present application are further described below in conjunction with the accompanying drawings.

[0030] Reference Figure 1 , Figure 1It is a structural diagram of a document analysis system based on a knowledge graph provided by an embodiment of the present application. The document analysis system is a comprehensive platform that integrates multiple core modules, specifically including a document knowledge graph construction module, an entity information extraction module, a query response module, and a data display module. Specifically, the document knowledge graph construction module can deeply mine key information such as entities, relationships, and attributes from a large amount of literature, and by integrating and associating this information, build a document knowledge graph with rich content and clear structure, providing a solid foundation for subsequent analysis and query. The entity information extraction module can accurately identify the query information input by the user and quickly lock the entities related to the query. Subsequently, the module will use these query entities to carry out efficient retrieval in the document knowledge graph and mine data information closely related to the query entity. The query response module can not only quickly generate a structured query list based on the retrieved data information, but also cleverly combine the document knowledge graph to provide users with comprehensive and accurate query results. This process not only reflects the intelligent processing capability of the system, but also highlights its efficient and accurate information retrieval and response capabilities. In addition, the data display module can provide users with intuitive and convenient query result visualization services. Among them, the multi-view visualization unit in this module can flexibly generate various types of visualization charts based on the query results, including but not limited to thematic word clouds, departmental donut charts, year stacked bar charts, and author tree charts. These charts are not only vivid, but can also intuitively reveal the key information and trends in the query results, providing users with a more intuitive and in-depth data insight and analysis experience.

[0031] In a possible embodiment, if Figure 1As shown in the figure, the literature knowledge graph construction module integrates a data preprocessing unit, a multi-dimensional processing unit and a graph generation unit. Among them, the data preprocessing unit can widely collect rich literature data from databases, document libraries and network resources, and clean and standardize them. This step aims to eliminate redundancy and errors in the data, ensure the accuracy, consistency and availability of the data, and thus generate high-quality standard literature data, laying a solid foundation for subsequent processing. The multi-dimensional processing unit can perform in-depth analysis of standard literature data, for example, using advanced text analysis technology to accurately extract key information such as entities, relationships and attributes from literature data. It should be noted that entities usually refer to specific objects mentioned in the literature, such as authors, research institutions, research topics, etc.; relationships describe the internal connections between entities, such as the cooperative relationship between authors, the citation relationship between documents, etc.; attributes are detailed descriptions of entity characteristics, such as the author's nationality, the publication year of the document and the keywords of the document. Based on the acquisition of this information, the multi-dimensional processing unit will construct a graph structure to present entities, relationships and attributes in an intuitive and clear manner. The graph generation unit is responsible for converting the graph structure into a literature knowledge graph. It can deeply analyze the characteristic information of the graph structure, such as the degree of nodes, the weight of edges, etc. These characteristic information not only reflects the structural characteristics of the literature knowledge graph, but also provides important clues for subsequent knowledge mining and application. Based on these characteristic information, the graph generation unit can use professional graph construction algorithms and technologies to construct a literature knowledge graph. This graph not only contains rich literature knowledge, but also shows the relationship and hierarchy between these knowledge in an intuitive and easy-to-understand way.

[0032] In a possible embodiment, if Figure 1 As shown in the figure, the entity information extraction module has a built-in retrieval enhancement unit, which can perform fine feature matching based on the feature information of the query entity and the graph structure, thereby obtaining accurate matching results. Based on these matching results, the module can accurately retrieve data information related to the query entity from a large amount of literature data, providing users with more accurate and targeted query results.

[0033] In a possible embodiment, if Figure 1 As shown in the figure, the document analysis system also integrates an interactive exploration module, which can perform real-time dynamic queries on visual charts based on various interactive operations input by users. This design not only ensures the instant and synchronous display of query results, but also greatly enhances the interactive experience between users and the system. Through this module, we can deeply explore the information hidden behind the charts, thereby achieving a comprehensive and in-depth understanding and analysis of document data.

[0034] See also Figure 2 , Figure 2It is the workflow diagram of the document analysis system provided by the embodiment of the present application. The system adopts a three-stage framework, and is intended to realize efficient analysis and visualization functions through dynamic queries generated based on multivariate metadata (i.e., detailed metadata of document data). In the first stage, the document knowledge graph construction module is responsible for processing extensive and complex multivariate document data, and generating corresponding graph representations (i.e., document knowledge graph). This module not only supports the flexible generation of dynamic queries, but also ensures the accuracy and pertinence of queries. By deeply mining the key information such as entities, relationships and attributes in the document data, and integrating and associating these information, a solid foundation can be laid for subsequent queries and analysis. In the second stage, the entity information extraction module works in conjunction with the query response module to conduct in-depth analysis of dynamic queries. These two modules make full use of the functions of advanced retrieval enhancement technology and large language models (LLMs) to accurately identify and efficiently retrieve query information. They not only retrieve data information closely related to the query entity from the document knowledge graph, but also conduct in-depth analysis and processing of these information, and finally generate query results with clear structure and rich content. It is worth noting that in the second stage, multivariate document data can be stored and integrated into a large language model as a database of the large language model, so that it can combine these multivariate document data and document knowledge graphs for comprehensive analysis and generate more accurate query results. In the third stage, the data display module visualizes these query results in an intuitive and easy-to-understand way. This module provides a variety of visualization charts and tools, such as subject word clouds, department pie charts, year stacked bar charts, and author tree charts. These charts not only vividly display the key information and trends in the query results, but also provide users with a deeper data insight and analysis experience. It is worth noting that users can perform rich interactive operations through these visualization charts and tools to achieve real-time dynamic queries on these charts. For example, users only need to simply click on a node in the bar chart to quickly query the details of the documents published under the department. This interactive visualization display method not only greatly improves the user experience, but also makes the data analysis process more flexible and efficient.

[0035] In a feasible embodiment, the process of constructing the literature knowledge graph module is as follows: Figure 3 As shown in , it is a process that comprehensively displays all levels of the knowledge graph by constructing multi-dimensional metadata. The starting point of the whole process is the collection of raw data, which is widely sourced from various literature materials, such as scientific publications. A typical scientific document may contain multiple entities, such as institutions, departments (such as department A, department B, department C) and authors (such as author A1, author A2 and author A3 from department A, author B1, author B2 and author B3 from department B, author C1, author C2 and author C3 from department C), etc. The internal connections between these entities are as follows: Figure 3 As shown. After obtaining the raw data, data preprocessing is performed to extract valuable metadata. This link covers multiple steps such as data extraction, cleaning, conversion and integration. Among them, data cleaning can identify and correct errors, omissions and duplications in the data to ensure the accuracy and consistency of the data; data conversion can convert data in different formats into a unified analyzable format, such as converting text data into the format required for natural language processing; data integration can integrate the cleaned and converted data into a complete data set. After the above processing, the raw data is converted into diversified metadata, which provides a solid foundation for the construction of the knowledge graph. Figure 3 In the paper, the document data is processed to form a clear data set, which classifies entities according to multiple attributes such as article name, author, year, and publication place. After obtaining diversified metadata, the next key step is to build a knowledge graph. This includes knowledge extraction, knowledge fusion (entity alignment), and data model construction. Knowledge extraction aims to extract entities, attributes, and relationships from metadata; knowledge fusion integrates the extracted information to eliminate duplication and contradiction to form a coherent knowledge network; data model construction is to build a suitable data structure to show the organization, interrelationships, and intrinsic characteristics of the data according to the application requirements of the knowledge graph. The constructed knowledge graph can provide a comprehensive overview from multiple dimensions, such as author, year, publication place, and department. Taking the department dimension as an example, the authors under the department, the articles they wrote, the publication place and year of the articles, etc. can be associated. The other dimensions are associated in a similar way to the department dimension, which will not be repeated here. By combining different dimensions and visualization techniques, the knowledge graph can provide users with richer, more intuitive, and more in-depth knowledge representation and analysis results, thereby helping users better understand and explore document knowledge.

[0036] See also Figure 4 , Figure 4It is a schematic diagram of the visualization interface of the document analysis system provided in the embodiment of the present application. This interactive visualization interface constructs three core components, aiming to provide users with a comprehensive and in-depth analysis experience. Specifically including: (A) Multi-angle view: This component aims to present data from a diversified perspective, allowing users to gain a comprehensive insight into the information behind the data and lay a solid foundation for in-depth analysis. Specifically, the views provided by this component include: a pie chart of the proportion of each department, which intuitively shows the proportion of each department in the whole; a straight line chart of the number of publications published by each department in different years, which clearly depicts the trend over time; a matrix chart of each department presents the relationship and difference between departments; and a word cloud view showing multiple keywords of the document data, emphasizing the importance of different data information in the overall analysis. These diverse view forms not only enrich the way data is presented, but also greatly enhance the user's understanding and analysis of data. (B) Knowledge graph display area: This area displays the corresponding knowledge graph through intuitive visualization means. By hovering over any node, the system will immediately display detailed information related to the node. For example, when hovering over an article node, the interface will clearly display key information such as the article title, list of authors, publication date, and publishing platform, greatly improving the convenience and intuitiveness of information acquisition. (C) Analysis dialogue module: This module integrates the advantages of retrieval enhancement technology and combines the intelligence of large language models with the rich knowledge base of integrated knowledge graphs to provide users with interactive analysis functions based on questions and answers. This design not only improves the accuracy of the analysis, but also greatly enhances the interactivity between users and the system. The three major components work together and complement each other to jointly build a system that supports multi-dimensional and in-depth analysis.

[0037] The knowledge graph-based document analysis system of this application is explained below with two application scenarios.

[0038] Application scenario 1: User A’s research interest exploration process

[0039] Assume that user A is an undergraduate who is preparing for the application of graduate school. In order to locate his personal research interests and find a suitable mentor, he decides to choose the literature analysis system of this embodiment to help him confirm his research direction. After carefully browsing the multi-dimensional view of the system, user A began to analyze the status of paper publication in each department in depth, and the data of department 2 particularly attracted his attention. Although the number of authors in this department is not large, its annual publication ratio is significantly high, which stimulated his further exploration behavior. With curiosity about the latest research direction of department 2, user A can ask a query in the analysis dialogue module: "In 2024, what research directions will department 2 focus on?" The system reveals that virtual reality is the current core research field of department 2 through the precise feedback of knowledge graph and related text analysis. This discovery gave user A a preliminary interest in virtual reality and prompted him to further look for authoritative experts in this field. Therefore, he can issue a second query in the analysis dialogue module: "The number of publications related to virtual reality and who are the authors?" The system responded quickly and listed a number of research papers related to virtual reality and their author lists. User A conducted a detailed investigation of the academic background and research direction of these experts. In the process of in-depth understanding, user A has a strong interest in the potential application of virtual reality in the metaverse. To this end, he sent a new request to the system's analysis dialogue module: "Please provide me with a list of relevant papers on the application of virtual reality in the metaverse." Based on his request, the system accurately pushed a series of papers closely related to virtual reality and the metaverse, covering a variety of application methods and scenarios. In order to understand the application scope of virtual reality more comprehensively and deeply, user A can continue to ask: "What are the popular application areas of virtual reality?" The system then showed a wealth of application cases of virtual reality in multiple fields such as games, education, medical care, and engineering design. These vivid examples greatly broadened user A's horizons and made him more clearly aware of the wide application value of virtual reality. Through this series of orderly exploration and inquiry, user A not only gained a comprehensive perspective on virtual reality, but also provided a valuable reference for his future research direction and mentor selection.

[0040] Application scenario 2: User B’s department analysis process

[0041] While reviewing the data of each department in the visualization interface, user B noticed that the number of publications of department 3 increased significantly between 2022 and 2023. To explore the root cause of this change, he tapped the data segment of department 3 in 2023 in the bar chart. This interaction activated the nodes and relationship network closely connected to it in the knowledge graph, presenting him with a more detailed association map. Then, the system intelligently constructed and executed an analytical query: "Explore the reasons for the surge in the number of papers in department 3 in 2023 and identify the professors who published the most papers." The analysis dialog box of the analysis dialogue module quickly fed back the results, and the knowledge graph was updated accordingly, highlighting the associated nodes of that professor. At the same time, the multidimensional view dynamically generated charts reflecting the latest situation based on this new information. When carefully examining the word cloud view, user B noticed that terms (keywords) such as "deep learning" and "human-computer collaboration" appeared frequently, suggesting that these may be the research focus of the professor. To further understand the professor's specific research achievements in these two fields, user B can construct a query: "Please summarize the professor's research contributions in the fields of 'deep learning' and 'human-computer collaboration'." The system responds immediately, and the knowledge graph and multidimensional view are updated synchronously again, complementing the in-depth analysis in the analysis dialog box. Through this series of carefully designed queries and in-depth analysis, user B can not only gain insight into the development context of Department 3, but also have a comprehensive and in-depth understanding of the professor's outstanding research work in the fields of deep learning and human-computer collaboration. This provides strong support for him to better evaluate the performance and future trends of the department.

[0042] See also Figure 5 , Figure 5 5 is a flow chart of a document analysis method based on a knowledge graph provided in an embodiment of the present application. The document analysis method can be applied to a document analysis system based on a knowledge graph as in the previous embodiment. The specific process of the method includes but is not limited to steps 510 to 550.

[0043] Step 510: extracting entities, relationships, and attributes from the document, and integrating the extracted entities, relationships, and attributes to construct a document knowledge graph;

[0044] Step 520: performing entity recognition on the query information input by the user to obtain query entities related to the query information;

[0045] Step 530: Retrieve data information related to the query entity from the literature knowledge graph based on the query entity;

[0046] Step 540: Generate a structured query list according to the data information related to the query entity, and generate a query result according to the structured query list and the literature knowledge graph;

[0047] Step 550: Visually display the query results.

[0048] In a feasible embodiment, the specific process of step 510 includes: collecting document data from databases, document libraries and network resources, and performing cleaning and standardization processing to obtain standard document data; parsing the text information in the standard document data to obtain entities, relationships and attributes, and constructing a graph structure based on the entities, relationships and attributes; extracting feature information of the graph structure, and constructing a document knowledge graph based on the feature information.

[0049] In a feasible embodiment, the specific process of step 530 includes: performing feature matching based on the feature information of the query entity and the graph structure to obtain a matching result; and querying the document data based on the matching result to obtain data information related to the query entity.

[0050] In a feasible embodiment, the specific process of step 550 includes: generating multiple visualization charts based on the query results and displaying the visualization charts, the multiple visualization charts including at least one of a keyword cloud, a department pie chart, a year stacked bar chart, and an author tree chart.

[0051] It should be noted that since the knowledge graph-based document analysis method of this embodiment can be applied to the knowledge graph-based document analysis system of the previous embodiment, the document analysis method of this embodiment and the knowledge graph-based document analysis system of the previous embodiment have the same technical principles and the same beneficial effects. In order to avoid repetition of content, they will not be repeated here.

[0052] Reference Figure 6 The embodiment of the present application further discloses an electronic device, the electronic device 600 comprising:

[0053] at least one processor 601;

[0054] At least one memory 602, used to store at least one program;

[0055] When at least one program is executed by at least one processor 601, the knowledge graph-based document analysis method described above is implemented.

[0056] An embodiment of the present application also discloses a computer-readable storage medium, which stores a computer program executable by a processor. When the computer program executable by the processor is executed by the processor, it is used to implement the knowledge graph-based document analysis method as described above.

[0057] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A document analysis system based on knowledge graph, characterized in that: include: A document knowledge graph construction module is used to extract entities, relationships and attributes from documents, and integrate the entities, relationships and attributes to construct a document knowledge graph; An entity information extraction module is used to perform entity recognition on the input query information to obtain query entities related to the query information, and retrieve data information related to the query entity from the literature knowledge graph based on the query entity; The query response module is used to generate a structured query list according to the data information, and generate a query result according to the structured query list and the literature knowledge graph.

2. The document analysis system according to claim 1, characterized in that: The literature knowledge graph construction module includes: The data preprocessing unit is used to collect literature data from databases, literature libraries and network resources, and to perform cleaning and standardization processing to obtain standard literature data; A multi-dimensional processing unit, used for parsing text information in the standard document data, obtaining entities, relationships and attributes, and constructing a graph structure based on the entities, the relationships and the attributes; The graph generation unit is used to extract feature information of the graph structure and construct a document knowledge graph based on the feature information.

3. The document analysis system according to claim 2, characterized in that: The entity information extraction module comprises: The retrieval enhancement unit is used to perform feature matching based on the feature information of the query entity and the graph structure to obtain a matching result, and based on the matching result, query the document data to obtain data information related to the query entity.

4. The document analysis system according to claim 1, characterized in that: The system further includes a data display module, which includes: A multi-view visualization unit is used to generate a variety of visualization charts based on the query results and display the visualization charts, wherein the multiple visualization charts include at least one of a subject word cloud, a department pie chart, a year stacked bar chart, and an author tree chart.

5. The document analysis system according to claim 4, characterized in that: The system further comprises: The interactive exploration module is used to dynamically query the visualization chart according to multiple interactive operations input.

6. A document analysis method based on knowledge graph, characterized in that: The following steps are involved: Extract entities, relationships and attributes from the document, and integrate the entities, relationships and attributes to construct a document knowledge graph; Performing entity recognition on the input query information to obtain query entities related to the query information; Retrieving data information related to the query entity from the literature knowledge graph based on the query entity; A structured query list is generated according to the data information, and a query result is generated according to the structured query list and the literature knowledge graph.

7. The document analysis method according to claim 6, characterized in that: The extracting entities, relations and attributes from the document and integrating the entities, relations and attributes to construct a document knowledge graph includes: Collecting literature data from multiple sources, and performing cleaning and standardization processing to obtain standard literature data, wherein the multiple sources include databases, literature libraries and network resources; Parsing text information in the standard document data to obtain entities, relationships and attributes, and constructing a graph structure based on the entities, the relationships and the attributes; Feature information of the graph structure is extracted, and a literature knowledge graph is constructed based on the feature information.

8. The document analysis method according to claim 7, characterized in that: The retrieving data information related to the query entity from the document knowledge graph based on the query entity includes: Perform feature matching based on the query entity and the feature information of the graph structure to obtain a matching result; Based on the matching result, the document data is queried to obtain data information related to the query entity.

9. An electronic device, characterized in that: include: at least one processor; at least one memory for storing at least one program; When at least one of the programs is executed by at least one of the processors, the document analysis method according to any one of claims 6 to 8 is implemented.

10. A computer-readable storage medium, characterized in that: A computer program executable by a processor is stored therein, and when the computer program executable by the processor is executed by the processor, it is used to implement the document analysis method as described in any one of claims 6 to 8.

Images