Research On The Theme Discovery And Evolution Of Domestic Library And Information Science Research Based On LDA

In the information age of the 21 st century the library and information science is full of unknown opportunities and challenges.With the rapid development of computer technology and Internet technology and the speed of document publishing,the research results in the field of library and information have exploded and the theme is more diverse.The topic model helps us discover and refine hidden,positive,and analyzable knowledge from the message text.LDA(Latent Dirichlet Allocation)is one of the most widely used probabilistic topic models.It is a three-layer Bayesian probability model consisting of three layers of words,topics and documents.Through the use of the Bag Of Word method,complex text information is transformed into mathematical information that is easy to process.At present,the field of library and information focuses on the application of this model for scientific literature topic mining and subject evolution research.In this paper,the LDA model is used to discover the research literature of the 12 years of domestic library and information science from 2006 to 2017.Firstly,the degree of confusion is used to determine the number of models,and the theme is identified according to the topic-term probability distribution file.Secondly,the topic strength of each topic is calculated according to the document-topic probability distribution file,and the current domestic library and information research topics are analyzed.Finally,the time factor is introduced,and the theme evolution trend is analyzed according to the topic intensity distribution,in order to provide data support and reference for the related research of library and information science in China.In the China Knowledge Network(CNKI)academic journal database,we obtained the abstracts of 10 core journals published in the field of library and information from 2006 to 2017.Using LDA to model,we found 20 books and information science research topics,namely,information literacy education for college students,theoretical research,evaluation research,librarian research,digital library and intellectual property,competitive intelligence,information organization,mobile library,knowledge management,public library and government information disclosure,user research,knowledge discovery,university library,big data,information resource construction,resource sharing,information services,information retrieval,online public opinion,reading promotion.The subject of the document is divided according to time,and the subject-vocabulary probability distribution and the document-topic probability distribution are obtained by using the post-discrete method.The calculation formula is used to obtain the intensity distribution of each topic,and the subject topic intensity time series is constructed.The research topics that found that the intensity of the topic is on the rise are university students' information literacy education,evaluation research,mobile library,user research,knowledge discovery,big data,online public opinion,reading promotion;the research theme with the theme strength decreasing trend is librarian research,Digital library and intellectual property,information organization,public library and government information disclosure,knowledge management,information resource construction,resource sharing,information service,information retrieval;research topics with small changes in subject intensity are theoretical research,competitive intelligence,University Library.