Research Topic Recommendation Based on Latent Dirichlet Allocation

Knowledge discovery pushes every discipline, including bibliometrics and innovation management, to reposition itself and stay competitive. With the exponential growth of scientific literature, it has become increasingly difficult for researchers to quickly and effectively identify potential research topics in an interested area to advance their research. This has led to the development of research topic recommendation. In this paper, aiming to explore the potential author-topic relations in depth, we proposed a research topic recommendation methodology based on latent Dirichlet allocation and bipartite network analysis, to predict the inspiring research topics that an author potentially will be interested and working on. A case study on scientific literature in the area of knowledge discovery is then presented to demonstrate the feasibility of the methodology. The result of this research could assist academic researchers in identifying and extending potential research directions.