Coal enterprise knowledge management system construction system and method

Abstract

The invention discloses a coal enterprise knowledge management system construction system and method in the technical field of computer information. The method comprises the following steps: collecting coal enterprise knowledge data; preprocessing the coal enterprise knowledge data to obtain knowledge graph data and large model module data; after performing data entity identification on the knowledge graph data by using an entity identification technology, creating a knowledge graph module; after large model module data are classified and labeled, a large model module is created; and creating a coal enterprise knowledge management system based on the knowledge graph module and the large model module. Through a mode of fusing the knowledge graph and the large model, the knowledge base answering field is enriched, the knowledge base question answering mode is optimized, the large model is enhanced by using the professional knowledge graph knowledge, the professional question answering result is generated, the question answering ability of the knowledge base in the professional field is improved, and contribution is made to digital, intelligent and green development of the coal industry.

Description

Technical Field

[0001] The invention relates to a coal enterprise knowledge management system construction system and method, belonging to the field of computer information technology. Background Art

[0002] The coal enterprise knowledge management system is a system or platform that centrally stores, manages and displays coal-related knowledge. It covers all aspects of coal, including but not limited to coal geological origin, mining technology, processing and utilization, market conditions, environmental impact, and relevant policies and regulations.

[0003] In the context of the rapid development of digitalization and intelligence in the coal industry, the management and application of industry knowledge has become a core problem. The existing knowledge management system lacks depth and professionalism, has poor relevance, and cannot achieve multi-dimensional analysis and application: the coal industry involves many professional technologies and complex issues that require in-depth research and understanding. However, some coal knowledge bases may simply list some basic concepts and common sense, lack in-depth analysis and interpretation, and cannot provide users with valuable professional guidance. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a system and method for constructing a knowledge management system for a coal enterprise. In view of the current situation of knowledge sedimentation and solidification in the existing knowledge management system of the coal industry, the system enriches the knowledge base answer field and optimizes the knowledge base question answering method by integrating knowledge graphs and large models, thereby contributing to the digitalization, intelligence and green development of the coal industry.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions: In a first aspect, the present invention provides a method for constructing a coal enterprise knowledge management system, comprising: Collect knowledge data of coal enterprises; Preprocess the knowledge data of coal enterprises to obtain knowledge graph data and large model module data; After using entity recognition technology to identify data entities in the knowledge graph data, create a knowledge graph module; After classifying and labeling the large model module data, create the large model module; Create a coal enterprise knowledge management system based on the knowledge graph module and the big model module.

[0006] Furthermore, the coal enterprise knowledge data includes internal data and external data: the internal data is provided by the enterprise's internal system, including structured data stored in the database and unstructured data not stored in the database; the external data is obtained through network collection and submission from other coal enterprises.

[0007] Furthermore, the coal enterprise knowledge data is preprocessed, including: using big data technology to clean, verify and classify the coal enterprise knowledge data, eliminating invalid data, abnormal data and untrue data, and then converting the unstructured data into structured data.

[0008] Furthermore, entity recognition technology is used to perform data entity recognition on knowledge graph data, including: using the MacBERT model as the word embedding layer, adding a long short-term memory network to extract the contextual semantic features of the knowledge graph data, and finally outputting the sum of the transfer score matrix and the observation score matrix of each word through the conditional random field, and using the Viterbi decoding algorithm to calculate the sequence with the highest cumulative score of all words, and then using entity fusion technology to merge two words with different expressions but the same actual meaning, and then using the Neo4j graph database to store the triple data of the knowledge graph.

[0009] Furthermore, the large model module parses the triple data of the knowledge graph by defining a knowledge template to generate question-answer pair data, and then rewrites the question-answer pair using the large model, and constructs a question-answer pair template using prompt word engineering technology, sets the maximum number of tokens and temperature variables of the large model, sends the question-answer pair template to the large model, automatically extracts question-answer pairs from the original semi-structured and unstructured text, performs data enhancement, and uses the knowledge graph content to infer questions that cannot be queried in the knowledge graph module, and scores the inference results. When the score obtained by the inference result is higher than the specified score, the result is used for question answering; After the large model module is created, the GLM open source pre-trained model is selected as the original base model, and then the question and answer data is used for training and fine-tuning, including: deduplication of the question and answer data, cleaning of irregular data operations, and converting it into the data set format required for fine-tuning the specified model; then the LoRA fine-tuning technology is used to adjust the specified parameters, including the rank of LoRA, the scaling factor of LoRA, and the dropout probability used in the LoRA layer; finally, the transformers library is used to fine-tune the model, including model loading, data pre-division and batching, optimizer and loss function definition, training parameter setting, and intermediate step model weight preservation; after the LoRA fine-tuning is completed, the model weights trained by LoRA need to be merged with the original base model.

[0010] Furthermore, the quality characteristics of the knowledge graph are evaluated based on a quality assessment method based on data quality vocabulary specifications, including: introducing additional abstract classes and attributes, constructing an ontology specifically for the quality of the knowledge graph in the coal mining field based on the data characteristics in the coal mining field, and then combining this ontology with the specific quality requirements, different quality dimensions, and quality categories of the knowledge graph, and concretizing the abstract classes and attributes therein to form a knowledge graph quality assessment ontology that meets specific needs, and finally using this ontology to comprehensively evaluate the intrinsic quality, contextual quality, description quality, and access quality of the knowledge graph, and supplement and improve the knowledge graph data based on the evaluation results.

[0011] Furthermore, the coal enterprise knowledge management system also includes a user portrait module and a software development module, wherein the user portrait module uses a clustering algorithm to initially create a user portrait, and then updates the user portrait in real time according to the user's software usage behavior to recommend personalized knowledge to the user; the software development module includes a system management module, a knowledge question and answer module and a security management module, wherein the system management module manages user data, enterprise data and authority data, the knowledge question and answer module answers questions input by users based on knowledge graphs and large models, and the security management module provides different data security protection solutions for static data and dynamic data.

[0012] In a second aspect, the present invention provides a coal enterprise knowledge management system construction system, comprising: Data collection module: used to collect knowledge data of coal enterprises; Data preprocessing module: used to preprocess the knowledge data of coal enterprises to obtain knowledge graph data and large model module data; Knowledge graph data processing module: used to create a knowledge graph module after using entity recognition technology to perform data entity recognition on the knowledge graph data; Large model module data processing module: used to create a large model module after classifying and labeling the large model module data; System building module: used to build a coal enterprise knowledge management system based on the knowledge graph after training and fine-tuning.

[0013] In a third aspect, the present invention provides a device for constructing a knowledge management system for a coal enterprise, including a processor and a storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of any of the methods described above.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of any of the above methods when executed by a processor.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention establishes a coal enterprise knowledge management system by integrating knowledge graphs and big models, enriches the answering fields of the knowledge base, optimizes the way of answering questions in the knowledge base, enhances the big model with professional knowledge graph knowledge, generates professional question-answering results, and improves the question-answering ability of the knowledge base in professional fields. At the same time, the reasoning ability of the big model is used to infer and score questions that cannot be queried in the knowledge graph, solving the problem of the illusion of answering questions by the big model. The data collection method of the present invention is extensive and targeted. In order to solve the problems of inaccurate knowledge sources and content in traditional knowledge bases, based on coal mine literature, laws and regulations, standards and specifications, encyclopedias and other unstructured text data, entities are extracted by establishing an entity recognition model, and the entity recognition model is evaluated to ensure the accuracy of the recognition and extraction content and to construct a coal mine field knowledge graph. A quality assessment method based on data quality vocabulary specifications is applied to evaluate the quality of the constructed graph, and the knowledge graph is supplemented and improved according to the evaluation results to improve the accuracy and comprehensiveness of the knowledge in the knowledge base. 3. The present invention is oriented to serve lean production, assist employees in daily production work, improve work efficiency, improve safety level, improve individual benefits, guide accurate decision-making analysis, and contribute to the digitalization, intelligence, and green development of the coal industry. By creating user portraits, personalized knowledge base related knowledge is pushed. Collect user information, adjust user portraits and push information in real time. Increase user interactivity and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings: Figure 1 A flowchart of a method for constructing a coal enterprise knowledge management system provided in the first embodiment of the present invention; Figure 2 A knowledge query flow chart of a method for constructing a coal enterprise knowledge management system provided in the first embodiment of the present invention; Figure 3 A safety risk classification diagram of a method for constructing a coal enterprise knowledge management system provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0017] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present invention can be combined with each other without conflict.

[0018] The following detailed description is an exemplary description, which is intended to provide further detailed description of the present invention. Unless otherwise specified, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present invention belongs. The terms used in the present invention are only for describing specific embodiments, and are not intended to limit the exemplary embodiments according to the present invention.

[0019] Embodiment 1: See also Figure 1 This embodiment provides a method for constructing a knowledge management system for a coal enterprise. The implementation method includes six steps: data collection, data processing, knowledge graph creation and quality assessment, large model fine-tuning, user portrait creation, and software development. Specifically, the method includes: Step 1, data collection is mainly used for coal enterprise knowledge collection. Coal enterprise knowledge is divided into internal data and external data. Internal data is mainly provided by the internal system of the enterprise, such as linking the internal database of the enterprise through an interface. Internal data includes: structured data stored in the database such as enterprise ERP, OA, design collaboration, etc.; unstructured data such as files, drawings, documents, scans, etc. External data is mainly obtained through network collection and submission by other coal enterprises, such as regular crawling and supplementation of network data through Scrapy crawler technology. External data includes: national standards, industry standards, group standards related to coal enterprises; coal papers and patents; coal-related enterprise information; coal-related industry terms and other data.

[0020] Step 2: Data processing mainly collects unstructured data, semi-structured data, and structured data for targeted processing. Then, knowledge extraction and entity fusion are performed on the knowledge graph module, and classification and labeling are performed on the large model module. Data processing is divided into three parts: data preprocessing, knowledge graph data processing, and large model data processing. Among them: Data preprocessing includes: using big data technologies such as decision tree algorithms, K-Means algorithms, and ETL tools to clean, verify, and classify data, and remove invalid data, abnormal data, and untrue data. Then, unstructured data is converted into structured data through technologies such as OCR and content recognition.

[0021] The data processing of knowledge graph includes: for the unstructured data and semi-structured data of the knowledge graph module, firstly, entity recognition technology is used to perform data entity recognition, in order to solve the problems of structured processing and key data extraction of paper materials such as design drawings, geological data, and hydrological data. The MacBERT model is used as the word embedding layer, and the long short-term memory network (LSTM) is added to extract contextual semantic features. Finally, the sum of the transfer score matrix and the observation score matrix of each word is output through the conditional random field (CRF), and the Viterbi decoding algorithm is used to calculate the optimal annotation sequence. The optimal sequence is the sequence with the highest cumulative score of all words. Then, the entity fusion technology is used to merge two words with different expressions but the same actual meaning. After the knowledge fusion is completed, the knowledge representation of the triple set is obtained, and the Neo4j graph database is selected to store the coal knowledge graph. In order to ensure the accuracy of entity recognition results, the entity recognition model is evaluated using accuracy, recall and F1 measurement indicators. When the model evaluation indicators meet certain requirements, the model is used to perform entity recognition on unlabeled data, and the entities extracted by the entity recognition model are manually reviewed to ensure the accuracy of the results to the greatest extent.

[0022] When evaluating the quality characteristics of the knowledge graph, a quality assessment method based on the data quality vocabulary specification is used to construct the knowledge graph quality dimensions from multiple perspectives such as completeness, consistency, accuracy, and coverage, obtain the knowledge graph quality assessment ontology, establish the knowledge graph quality model, and evaluate the knowledge graph quality. Specifically, first, by introducing additional abstract classes and attributes on the basis of the existing data quality vocabulary, an ontology specifically for the quality of the knowledge graph in the coal mine field is constructed for the data characteristics of the coal mine field. Next, this ontology is combined with the specific quality requirements, different quality dimensions, and quality categories of the knowledge graph, and the abstract classes and attributes therein are concretized, thereby forming a knowledge graph quality assessment ontology that can meet specific needs. Finally, this ontology is used to comprehensively evaluate the endogenous quality, contextual quality, description quality, and access quality of the knowledge graph. The knowledge graph data is supplemented and improved based on the evaluation results.

[0023] For the data of the large model module, select some of the obtained data for effective classification and annotation. The large model module is mainly used for semantic recognition, understanding user input, and can call knowledge graph data, organize output data to generate text. When the knowledge graph data is not enough to answer user questions, the large model module can be used to perform deep reasoning on the knowledge graph module related knowledge, and use high-scoring answers to answer, in order to deal with the phenomenon of hallucination of answering questions by the large model. Specifically, the large model data processing includes three parts. The first is to define the knowledge template, parse the triple data of the knowledge graph, generate question-answer pairs, and then use the large model to rewrite the question-answer pairs for large model knowledge enhancement. The second is to use the prompt word engineering technology to construct the question-answer pair template, set the maximum number of tokens and temperature variables of the large model, send the template to the large model, and automatically extract the question-answer pairs from the original semi-structured and unstructured text for data enhancement. The third is to use the knowledge graph content to reason about questions that cannot be queried in the knowledge graph module. When the score obtained by the result is higher than the specified score, the result is used for question answering.

[0024] Step 3: Create a knowledge graph ontology, configure knowledge mapping, and store the knowledge graph data obtained in step 2 into the Neo4j graph database to create a knowledge graph. The knowledge graph mainly processes query requests for large model modules and outputs query results.

[0025] Step 4, select the latest open source pre-trained model of GLM as the original base model, and then use the question-answer pair data obtained in step 2 to train and fine-tune. First, deduplicate the question-answer pair data, clean the irregular data operations, and convert it into the data set format required for the specified model fine-tuning; then use the LoRA fine-tuning technology to adjust the specified parameters, including the rank of LoRA, the scaling factor of LoRA, and the dropout probability used in the LoRA layer; finally, use the transformers library to fine-tune the model, including model loading, data pre-division and batching, optimizer and loss function definition, training parameter setting, intermediate step model weight preservation, etc. After the LoRA fine-tuning is completed, the model weights trained by LoRA need to be merged with the original base model.

[0026] Step 5, software development is mainly divided into computer-side development and mobile-side development. The computer side is mainly displayed on the web page, the front end uses the VUE language, and the back end uses the JAVA language. The mobile phone side is displayed on the WeChat applet, the front end is developed using the WeChat applet development interface, and the back end is developed using JAVA. The software development functions include three modules: system management, knowledge question and answer, and security management. Among them, the system management module manages user data, enterprise data, and permission data. The knowledge question and answer module is the main functional module of the present invention, and users can access the knowledge graph and the big model in the form of questions and answers. The knowledge question and answer module is mainly based on the knowledge graph and the big model to answer questions entered by the user. The user's question is first answered by the knowledge graph. If there is an answer, it is further sorted out with the big model. Otherwise, the big model module is used to perform deep reasoning on the knowledge related to the knowledge graph module, and high-scoring answers are used to answer. Combine the knowledge graph and the big model to generate professional question and answer results.

[0027] A clustering algorithm is used to initially create a user portrait, which is then updated in real time based on the user's software usage behavior in order to recommend personalized knowledge to the user.

[0028] Access process as follows Figure 2 First, the knowledge query request is initiated at the front end and submitted to the back-end big model module for semantic recognition. Then the query is performed through the knowledge graph module. If the knowledge graph query finds a result, it is handed over to the big model module for summary and text and image output; if there is no query result in the knowledge graph, it is handed over to the big model to reason about the relevant data of the knowledge graph and answer it by itself. Finally, the answer result is output to the user display interface. This method can effectively ensure the accuracy of knowledge answers.

[0029] It should be noted that the coal enterprise knowledge management system can update user portraits based on user behavior. The system can periodically fine-tune the large model based on user questions and user feedback. It can also periodically update data to ensure the real-time nature of the data.

[0030] Please refer to Figure 3 , data security risks include static data security risks and dynamic data security risks. Static data security risks include personal data security risks and enterprise data security risks. Dynamic data security risks include data collection stage, model training stage, and model application stage. The security management module is used to protect the data security of the present invention and provide different data security protection solutions for static data and dynamic data. In terms of hardware, network security equipment can be used to manage networked devices. In terms of software, data security development can be used for data management. To achieve data security and stability. At the same time, users need to have permission control when querying data. Users with different permissions can only access data within the permission range.

[0031] Embodiment 2: A coal enterprise knowledge management system construction system can implement a coal enterprise knowledge management system construction method described in Example 1, including: Data collection module: used to collect knowledge data of coal enterprises; Data preprocessing module: used to preprocess the knowledge data of coal enterprises to obtain knowledge graph data and large model module data; Knowledge graph data processing module: used to create a knowledge graph module after using entity recognition technology to perform data entity recognition on the knowledge graph data; Large model module data processing module: used to create a large model module after classifying and labeling the large model module data; System building module: used to build a coal enterprise knowledge management system based on the knowledge graph after training and fine-tuning.

[0032] Embodiment three: The embodiment of the present invention further provides a coal enterprise knowledge management system construction device, which can implement a coal enterprise knowledge management system construction method described in the first embodiment, including a processor and a storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to perform the steps of the following method: Collect knowledge data of coal enterprises; Preprocess the knowledge data of coal enterprises to obtain knowledge graph data and large model module data; After using entity recognition technology to identify data entities in the knowledge graph data, create a knowledge graph module; After classifying and labeling the large model module data, create the large model module; Create a coal enterprise knowledge management system based on the knowledge graph module and the big model module.

[0033] Embodiment 4: The embodiment of the present invention further provides a computer-readable storage medium, which can implement the method for constructing a coal enterprise knowledge management system described in the first embodiment, and stores a computer program thereon, which implements the steps of the following method when the program is executed by a processor: Collect knowledge data of coal enterprises; Preprocess the knowledge data of coal enterprises to obtain knowledge graph data and large model module data; After using entity recognition technology to identify data entities in the knowledge graph data, create a knowledge graph module; After classifying and labeling the large model module data, create the large model module; Create a coal enterprise knowledge management system based on the knowledge graph module and the big model module.

[0034] It is known from common technical knowledge that the present invention can be implemented by other embodiments that do not deviate from its spirit or essential features. Therefore, the above disclosed embodiments are only illustrative in all respects and are not exclusive. All changes within the scope of the present invention or within the scope equivalent to the present invention are included in the present invention.

[0035] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0036] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram and the combination of processes and / or blocks in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0037] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0038] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0039] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for constructing a knowledge management system for a coal enterprise, characterized in that: include: Collect knowledge data of coal enterprises; Preprocess the knowledge data of coal enterprises to obtain knowledge graph data and large model module data; After using entity recognition technology to perform data entity recognition on the knowledge graph data, a knowledge graph module is created. The knowledge graph module uses the MacBERT model as the word embedding layer, and adds a long short-term memory network to extract the contextual semantic features of the knowledge graph data. Finally, the sum of the transfer score matrix and the observation score matrix of each word is output through the conditional random field, and the Viterbi decoding algorithm is used to calculate the sequence with the highest cumulative score of all words. Then, the entity fusion technology is used to merge two words with different expressions but the same actual meaning. After obtaining the triple data of the knowledge graph, the Neo4j graph database is selected for storage; After classifying and labeling the big model module data, create a big model module, the big model module parses the triple data of the knowledge graph by defining the knowledge template, generates question-answer pair data, and then rewrites the question-answer pair using the big model, and uses prompt word engineering technology to build a question-answer pair template, sets the maximum number of tokens and temperature variables of the big model, sends the question-answer pair template to the big model, automatically extracts question-answer pairs from the original semi-structured and unstructured text, performs data enhancement, and uses the knowledge graph content to infer questions that cannot be queried in the knowledge graph module, and scores the inference results. When the score obtained by the inference result is higher than the specified score, the result is used for question answering; Create a coal enterprise knowledge management system based on the knowledge graph module and the big model module.

2. The method for constructing a coal enterprise knowledge management system according to claim 1, characterized in that: The coal enterprise knowledge data includes internal data and external data: the internal data is provided by the enterprise's internal system, including structured data stored in the database and unstructured data not stored in the database; the external data is obtained through network collection and submission from other coal enterprises.

3. The method for constructing a coal enterprise knowledge management system according to claim 1, characterized in that: Preprocessing the coal enterprise knowledge data includes: using big data technology to clean, verify and classify the coal enterprise knowledge data, eliminating invalid data, abnormal data and untrue data, and then converting the unstructured data into structured data.

4. The method for constructing a coal enterprise knowledge management system according to claim 1, characterized in that: After the large model module is created, the GLM open source pre-trained model is selected as the original base model, and then the question and answer data is used for training and fine-tuning, including: deduplication of the question and answer data, cleaning of irregular data operations, and converting it into the data set format required for fine-tuning the specified model; then the LoRA fine-tuning technology is used to adjust the specified parameters, including the rank of LoRA, the scaling factor of LoRA, and the dropout probability used in the LoRA layer; finally, the transformers library is used to fine-tune the model, including model loading, data pre-division and batching, optimizer and loss function definition, training parameter setting, and intermediate step model weight preservation; after the LoRA fine-tuning is completed, the model weights trained by LoRA need to be merged with the original base model.

5. The method for constructing a coal enterprise knowledge management system according to claim 1, characterized in that: The quality characteristics of the knowledge graph are evaluated based on a quality assessment method based on data quality vocabulary specifications, including: introducing additional abstract classes and attributes, constructing an ontology specifically for the quality of the knowledge graph in the coal mining field based on the data characteristics in the coal mining field, and then combining this ontology with the specific quality requirements, different quality dimensions, and quality categories of the knowledge graph, and concretizing the abstract classes and attributes therein to form a knowledge graph quality assessment ontology that meets specific needs, and finally using this ontology to comprehensively evaluate the intrinsic quality, contextual quality, description quality, and access quality of the knowledge graph, and supplementing and improving the knowledge graph data based on the evaluation results.

6. The method for constructing a coal enterprise knowledge management system according to claim 1, characterized in that: The coal enterprise knowledge management system also includes a user portrait module and a software development module, wherein the user portrait module uses a clustering algorithm to initially create a user portrait, and then updates the user portrait in real time according to the user's software usage behavior, and recommends personalized knowledge to the user; The software development module includes a system management module, a knowledge question and answer module and a security management module, wherein the system management module manages user data, enterprise data and permission data, the knowledge question and answer module answers questions input by users based on knowledge graphs and large models, and the security management module provides different data security protection solutions for static data and dynamic data.

7. A coal enterprise knowledge management system construction system, characterized in that: include: Data collection module: used to collect knowledge data of coal enterprises; Data preprocessing module: used to preprocess the knowledge data of coal enterprises to obtain knowledge graph data and large model module data; Knowledge graph data processing module: used to create a knowledge graph module after using entity recognition technology to perform data entity recognition on the knowledge graph data. The knowledge graph module uses the MacBERT model as the word embedding layer, and adds a long short-term memory network to extract the contextual semantic features of the knowledge graph data. Finally, the conditional random field is used to output the sum of the transfer score matrix and the observation score matrix of each word, and the Viterbi decoding algorithm is used to calculate the sequence with the highest cumulative score of all words. Then, the entity fusion technology is used to merge two words with different expressions but the same actual meaning. After obtaining the triple data of the knowledge graph, the Neo4j graph database is selected for storage; Large model module data processing module: used to create a large model module after classifying and annotating the large model module data. The large model module parses the triple data of the knowledge graph by defining a knowledge template to generate question-answer pair data, and then rewrites the question-answer pair using the large model, and uses prompt word engineering technology to build a question-answer pair template, set the maximum number of tokens and temperature variables of the large model, send the question-answer pair template to the large model, automatically extract question-answer pairs from the original semi-structured and unstructured text, perform data enhancement, and use the knowledge graph content to infer questions that cannot be queried in the knowledge graph module, and score the inference results. When the score obtained by the inference result is higher than the specified score, the result is used for question answering; System building module: used to build a coal enterprise knowledge management system based on the knowledge graph after training and fine-tuning.

8. A coal enterprise knowledge management system construction device, characterized in that: including processor and storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 6 are implemented.

Images