Discovering the role model of authors by embedding research history

Abstract

A role model that supports career planning is important for authors in the academic area to improve research abilities. In this study, we discovered a role model in bibliographic networks based on two perspectives: (1) high research performance to be exemplary and (2) a similar research history that can be easily followed by authors. We assume that the year-wise subgraphs in the dynamic bibliographic network signify the 'research history'. We discovered role models of authors in three steps: (1) learning vector representations of research history in dynamic bibliographic networks, (2) measuring the similarity of authors according to the research history and (3) visualising role models. With this process, we can recommend a reasonable role model whose research path the authors can easily follow. In addition, we verified the effectiveness of the research history embeddings and the accuracy of the recommended role model in a real data set.