Abstract

Co-authorship in scientific research is increasing in the past decades. There are lots of researches focusing on the pattern of co-authorship by using social network analysis. However, most of them merely concentrated on the properties of graphs or networks rather than take the contribution of authors to publications and the semantic information of publications into consideration. In this paper, we employ a contribution index to weight word vectors generated from publications so as to represent authors’ research interest, and try to explore the relationship between research interest and co-authorship. Result of curve estimation indicates that research interest couldn’t be employed to predict co-authorship. Therefore, graph-based researcher recommendation needs further examination.

Share

COinS