Abstract: Timely and accurate grasp of research fronts can provide a more comprehensive decisionmaking basis and reference for the formulation of scientific and technological policies and scientific research deployment. In the era of big data, the effective and cooperative use of multi-source data to identify research fronts has become the research focus in the field of information science. Based on the literature review and content analysis, this paper analyzed the method of the research fronts identification and multi-source data fusion in depth. According to the different research objects, this paper divided the identification methods into citation-based, lexical-based, topic-based and fusion-based methods, compared the difference, and expounded the necessity of fusion-based methods. In terms of multi-source data fusion methods, this paper reviewed the characteristics and shortcomings of existing methods from carrier fusion and relationship fusion, according to the depth of fusion. We constructed a carrier-feature-relationship fusion model for multi-source scientific text fusion. What’s more, based on analyzing the core features of research fronts, we proposed three identifying indicators, namely, degree of attention, novelty and centrality. This study enriches method of the research fronts identification based on multi-source data fusion.

Key words: Multi-source data fusion, Carrier fusion, Feature fusion, Relationship fusion, Research fronts identification, Topic model