Research And Implementation Of Academic Knowledge Graph Based On Natural Language Processing Technology

In the era of data explosion,all industries are faced with the challenge of comprehensive digital transformation.How to mine the intrinsic relationship between data and visualize the relationship between data becomes the key.In the academic field,countless scholars publish many academic papers every year,and each academic website has a large amount of data in its own right.Although data silos are good for data maintenance and control,they will reduce data quality,data viewing speed and data analysis accuracy.At present,the mainstream method to check the status of teachers’ published papers,the cooperative relationship and citation relationship between teachers,and the details of papers is to obtain academic information through webpage display on academic websites.However,there are many cooperative relationships and citation relationships between teachers,and the web page display is not enough to allow viewers to establish a clear academic knowledge structure in their minds and intuitively understand the situation.As a research boom in recent years,knowledge graphs can add various associations and relationships to connect data islands and provide a complete and visible view of data query.Especially in the academic field,knowledge graphs can systematically display complex academic information,including scholar information and paper information,and change the original mode of displaying academic knowledge query results in web pages.Moreover,due to the characteristics of knowledge graph,academic knowledge query can be made simple and fast,and the academic relationship between teachers can be displayed clearly and intuitively.Therefore,this paper designs and implements a visual academic knowledge graph platform to solve the problem of structured display and integration of academic knowledge.The main work of this paper is as follows:The first is data source acquisition and preprocessing.This part mainly extracts the unstructured text information of teachers’ homepages on college websites,as well as structured and semi-structured academic data on academic websites including Baidu Academic and DBLP webpages.The second is the acquisition of academic knowledge.After defining the ontology,for the semi-structured data and structured data on the academic website,the Jsoup framework is directly used to crawl out the academic knowledge triples;for the unstructured text data,this paper explores the natural language processing-based Bi LSTM-CRF trains the model and uses it for entity extraction experiments.The entity relationship extraction link uses the syntactic dependency analysis method to extract the entity relationship from the text sentence.The knowledge representation and storage stage is completed by using the RDF triple representation method combined with the Neo4 j graph database.Finally,the knowledge triplet is used to realize the construction of the academic knowledge graph visualization platform.The underlying data module of the system uses graph database storage and query technology,the back-end uses the Spring Boot framework,and the frontend uses the Echarts tool to render the queried data,and finally displays it in the form of a "node-relation-node" relationship diagram.And the query module and information display module of the paper.