Construction And Application Of Knowledge Graph Oriented To Professional Literature

For scholars,the literature in the professional field is the data that they face most frequently in the daily academic research.In the era of big data for research,it is easy for scholars to fall into the quagmire when they face the millions of literature resources.As an emerging technology in the knowledge economy,the knowledge graph can describe entities and their relationships in a structured manner from a semantic perspective.Therefore,it is of great value for scholars to apply the knowledge graph to the academic field,and to describe the knowledge network.The paper designs the overall construction framework of the literature knowledge graph,taking the machine learning literature as an example to build the graph from the pattern layer and the instance layer.At last,relevant applications are carried out on the basis of literature knowledge graph.The research content is as follows.First,the paper improves the general construction process of knowledge graphs in vertical domains,and proposes a framework for constructing knowledge graph that is applicable to literature in professional domains.The framework contains the overall process of constructing the domain literature knowledge graph and the specific methods used in each step.Second,for the construction of the pattern layer,the paper adopts a bottom-up approach.By adjusting the term clustering results artificially and combining the needs of scholars,the four major entity types and the five major relationship types between entities are determined.This method makes up for the problem of over-reliance on domain expert knowledge in the current research.Third,for the construction of the instance layer,this paper uses long-short term memory networks and conditional random field model to realize the identification and labeling of knowledge entities.For entity relationship extraction,the paper introduces attention mechanism based on long-short term memory networks.Then,the paper verifies the validity of models through experiments.Fourth,based on the literature knowledge graph stored in graph database,two application examples—literature marking tool and knowledge discovery tool—are designed.These tools can realize the functions of literature marking,knowledge searching and formatted overview.The paper studies the construction and application of knowledge graph in the professional literature field.Theoretically,based on the adaptive improvement of the construction process about the vertical domain knowledge graph,the domain literature knowledge graph is completely constructed,which further enriches and develops the research of knowledge graph.In practice,through the knowledge graph application,the knowledge implicit in the literature will be displayed in an explicit manner,which will improve the efficiency of scholars' literature reading and help them understand the current research in the field more quickly.