A structural topic modeling-based bibliometric study of sentiment analysis literature

Sentiment analysis is an increasingly evolving field of research in computer science. With the considerable number of studies on innovative sentiment analysis available, it is worth the effort to present a review to understand the research on sentiment analysis comprehensively. This study aimed to investigate issues involved in sentiment analysis; for instance, (1) What types of research topics had been covered in sentiment analysis research? (2) How did the research topics evolve with time? (3) What were the topic distributions for major contributors? (4) How did major contributors collaborate in sentiment analysis research? Based on articles retrieved from the Web of Science, this study presented a bibliometric review of sentiment analysis with the basis of a structural topic modeling method to obtain an extensive overview of the research field. We also utilized methods such as regression analysis, geographic visualization, social network analysis, and the Mann–Kendal trend test. Sentiment analysis research had, overall, received a growing interest in academia. In addition, institutions and authors within the same countries/regions were liable to collaborate closely. Highly discussed topics were sentiment lexicons and knowledge bases, aspect-based sentiment analysis, and social network analysis. Several current and potential future directions, such as deep learning for natural language processing, web services, recommender systems and personalization, and education and social issues, were revealed. The findings provided a thorough understanding of the trends and topics regarding sentiment analysis, which could help in efficiently monitoring future research works and projects. Through this study, we proposed a framework for conducting a comprehensive bibliometric analysis. Copyright © 2020 Springer Science+Business Media, LLC, part of Springer Nature.