Research On Information Retrieval Discipline Topic Based On Comparative Study On SIGIR Mailing List And Scholar Papers

The discipline topic research mainly includes the research of discipline topic structure and research of discipline topic evolution.Most of the existing discipline topic research is based on the literature data of scholar papers.Although this method is effective,but it's too stereotyped.There are many scholars abroad have studied the mailing list of open source software,these research focus on the mining of mail content and the social network of mailing list.In the light of the related work,we construct mailing list dataset and scholar paper dataset based on the SIGIR mailing list.The SIGIR mailing list consists of 7218 mails from 2008 to 2016,and the conference papers dataset contains 1968 papers published on SIGIR and 828 papers publish on ECIR.In these mails,we find 1312 scholars in the ACM digital library according to the mail senders.First,we study the structure of discipline topic based on the two data sets:(1)We classify mailing list to conference announcement and job posting.We identify 40 frequent keywords using the TF-IDF algorithm,building keywords co-occurrence network on the three datasets respectively and clustering the keywords with SPSS.We find seven research topics of information retrieval respectively.(2)We build keyword-scholar co-occurrence network to visualize the relationship between keyword and scholar.(3)We conclude eight research topics,and introduce each topic respectively.Then,we study the evolution of discipline topics of information retrieval:(1)Analyzing the seven research topics timeline in recent 10 years using word frequency analysis in the three dataset respectively.(2)Finding the representative keywords of each year using keyword-year co-occurrence network.(3)Using NEViewer to visualizing the evolution of information retrieval discipline.