Detecting emerging topics by exploiting probability burst and association rule mining: A case study of Library and Information Science

Abstract

The primary reason for detecting emerging topics is to reduce researchers’ time in finding current related topic while maintaining awareness of current trends in a particular field. Nowadays, the amount of information is growing rapidly, but tracking the development of a research field by manually reading the literature is challenging. This study takes Library and Information Science (LIS) as a case study to present a new method for detecting emerging topics. This novel method could be applied to analyse various types of documents and detect emerging topics automatically. This method utilizes a Latent Dirichlet Allocation (LDA) model to generate topics and calculate probabilities. It discovers emerging topics by detecting probability burst in consecutive time spans. Association rule mining and lexical similarity computation are adopted to represent the topics. This work tests the method by comparing the results of emerging topics from the LIS data in the baseline paper. The validation demonstrates that the proposed approach is feasible.