[Purpose/significance] In recent years, the cooperation among scholars has become one of the important ways to output academic achievements.Co-authorship link prediction is one of the important topics to improve scientif? ic research efficiency and promote the integration of discipline resources. However, little studies have treated the coauthorship network as a dynamic time-varying network and integrated textual semantics and dynamic network informa? tion. [Method/process] Therefore, we propose a co-authorship link prediction model which incorporatenodes’unstruc? tured semantic attributes and dynamic network features.First, we take statisticians as an example and collect all papers in four top statistical journals from 2011 to 2020, which helps us construct a statisticians’co-authorship network. Sec? ond, we extract unstructured semantic attributes and dynamic network feature to enrich nodes representation. Finally, based on these features, we make a link prediction based on a dynamic logical regression model. [Result/conclusion] The results have shown that several factors can significantly influence co-authorship, such as whether two researchers cooperated last year, the research similarity between two researches, citation-related factors, etc. The result of this paper is of great significance to enhance the relationship between scholars and the efficiency of scientific research cooperation.