Research On The Application Of Intelligent Retrieval Based On Chinese Book Knowledge Graph

As the document information service center of colleges and universities,university library plays an important role in providing accurate knowledge service for teachers and students.According to the general requirements of "Education Informationization 2.0 Action Plan" and the strategic task of "China’s Education Modernization 2035",in the process of building digital campus,it is very important to explore the appropriate library precise knowledge service mode for the intelligent construction of libraries.As the main way to transfer the library resources,the knowledge service mode of university library resource retrieval platform not only affects the circulation rate of the library resources,but also plays an important role in the sustainable development of the library.However,most of the existing library resource retrieval platforms in colleges and universities use the user-independent query method of keyword matching,which is easy to lead users into the dilemma of "information overload" and "information trek",which greatly limits the intelligent service level of libraries.In view of the above problems,this paper takes book keywords as the entry point,makes full use of the advantages of knowledge graph technology in knowledge structure and relationship representation,carries out the research on the intelligent retrieval application of Chinese book knowledge graph with book keywords as the core,and explores the innovative application approaches of intelligent service in university libraries.The main research contents of this paper are as follows:(1)Research on the method of keyword extraction and generation integrating the attention of title and subject words.In this paper,the unsupervised method is used to extract and generate keywords as the keyword tags of books,and then the book keyword tags are used to train the deep learning model.Considering the importance of book titles and subject headings to books,the attention of book titles and subject headings is integrated in the process of training the model.Finally,the effectiveness of the model is verified through the experimental comparison on the self-built data set.(2)Construction of knowledge graph of Chinese books integrating discipline classification.This paper constructs the book ontology according to the tree structure of subject catalog and related concepts of books,integrates three knowledge triples of "concept-attribute-value","subject-subordinate subject-superior discipline" and "keyword-relationship-keyword" according to the ontology constraint and relational definition,and designs the corresponding entity alignment algorithm according to the relationship between keywords.Finally,Neo4 j is used to store the knowledge graph(SCBKG)of Chinese books integrating disciplines classification.(3)Research on the application of intelligent book retrieval based on SCBKG.Aiming at the two scenarios of user input search term and search sentence,the intelligent keyword search model based on SCBKG and the intelligent topic sentence search model based on SCBKG were designed respectively according to the differences of the scenarios.Finally,the reliability of the search model was verified through the self-built test set,and the necessity of integrating the knowledge graph in the search process was confirmed.(4)Design and development of intelligent retrieval system for Chinese books.Based on the constructed Chinese book knowledge graph and key technologies,Django framework is adopted to build the Chinese book intelligent retrieval system,and the book resource management and intelligent retrieval service based on the visualization of knowledge graph are realized.The system mainly includes search term retrieval,search sentence retrieval and other core function modules,which can not only satisfy users’ visual retrieval,but also adjust the scope of retrieval according to user interaction,so as to better realize the accurate retrieval service of book resources.