Literature Navigation And Expert Eecommendation Technology And Application For Knowledge Service

The ever-expanding online literature library gives researchers full access to the latest scientific research and experts and scholars.However,the sheer volume means the difficulty of choice,and today it has gradually become a major difficulty in scientific research.On the one hand,the large number of online literature databases allows researchers to experience problems such as information overload,learning to navigate and learning backtracking during the research and learning process.On the other hand,in order to save research time,it is an efficient work mode to select potential partners with appropriate professional knowledge for joint learning.However,how to find the most suitable collaborators from experts and scholars around the world is still a great problem.Aiming at these two problems,this thesis takes the literature knowledge base as the research object.Based on the literature network structure,this thesis proposes a knowledge map and learning path recommendation algorithm based on the optimal selection of associated literature and an expert recommendation technology based on network embedding algorithm.It is expected that the above two problems will be solved from the two aspects of planning literature learning order and expert recommendation.The specific research work includes the following three parts:(1)A method of constructing a literature knowledge map is proposed.Based on the citation relationship between the literature data,this thesis designs a contribution conduction between document nodes and a literature value calculation method,which can select high-quality literature to build literature knowledge maps,realize the recommendation of learning paths,and plan the order of literature learning.The experimental results prove that the recommended literature and its learning order are scientific.(2)An expert recommendation technique based on network embedding algorithm(Fusing High-Order Proximity of Structure and Content via Nonnegative Matrix Factorization for Embedding Information Networks,FSCEN)is proposed.The adjacency matrix of the network structure and the content matrix of the network node are respectively constructed,and the best non-negative matrix decomposition is calculated by using the consistency between the structure and the content to realize the feature dimension reduction,and the spectral clustering completes the clustering of the network nodes.Calculate expert impact values in each category,combined with the activity of the experts over the years to achieve expert list recommendations.(3)The scientific research navigation system was built.The system architecture,and sub-module are introduced in detail to introduce the functions and specific implementation process of the system.Finally,the view display is completed in the form of a web page.