A Comparative Study on Short-Text Topic Modeling in Online Collaborative Learning Platforms: Application and Performance Analysis of BTM and BERTopic
Year: 2024
Type: conference-paper
Abstract: This paper addresses the challenges posed by the sparse semantics and complexity of short text data in collaborative e-learning platforms by using two topic modelling methods, BTM (Biterm Topic Model) and BERTopic. Through data pre-processing, topic modelling and visualisation, we systematically compare the applicability of these two models in different data scenarios. The results show that the BERTopic model excels in processing complex data containing both long and short texts, effectively capturing deeper semantic information, while the BTM model offers better interpretability and stability in smaller data sets. This study offers new methodological support for text analysis in e-learning platforms and provides indications for future pedagogical feedback and support for personalised teaching.
Authors: Kening Zhu, Jidong Weng, Binghao Tu
Institution: Zhejiang University of Technology
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: gold
OpenAlex ID: W4409965715