Identification of Topic Trends in the Journal of Information Science (2015-2024) Using a Cluster Embedding Technique

This study aims to identify the trend of research topics in the Journal of Information Science from 2015 to 2024 using bibliographic records retrieved from the Dimensions database. A total of 819 records were collected by filtering for the journal’s name. BERTopic, a cluster embedding technique, was employed for topic modeling on the abstracts of the records. Three machine learning models–Naïve Bayes, Support Vector Machine (SVM), and Random Forest were used for classifying topics. The results show that 2022 had the highest number of publications (n=142). This study also identified twelve distinct fields of research, with “Information and Computing Sciences” being the most prominent. Topic modeling revealed five research topics: “Information Management,” “Bibliometrics,” “Text Classification,” “Semantic Web,” and “Recommender Systems.” In terms of classification, SVM achieved the highest accuracy (85%) and outperformed the other two models. These findings provide valuable information about the evolving research landscape in the field of information science and highlight the increasing focus on the five research topics. This study enhances the understanding of publication trends and provides a methodological framework for future research.