Adoption of Various Topic Modelling Algorithm to Analysis Indonesian Tourism Customer Feed Back
Muhammad Hanafi, Icha Nura Nugraha, Sumarni Adi · 2024
The creative economy and tourism industry are significant sources of foreign exchange and major contributors to a country's gross domestic product (GDP). This study evaluates trends in the creative economy and tourism industry using topic modeling approaches, namely Latent Dirichlet Allocation (LDA), Latent Semantic Analysis (LSA), and BERTopic. Data were collected from the Twitter platform using API and analyzed with preprocessing steps including tokenization, stemming, and lemmatization. Model evaluation was performed based on topic coherence metrics (UMass, UCI, and Cv) and processing time The results show that LDA and LSA have high stability and consistency in generating coherent topics, while BERTopic has great potential but requires further calibration to achieve comparable stability. This study provides insights into the most effective topic models for analyzing trends in the creative economy and tourism industry, as well as providing recommendations for future research.