Automatic Topic Modeling Using Optimized BERTopic
13th International Conference on Electrical and Electronics Engineering, ICEEE 2026, Antalya, Türkiye, 27 - 29 Nisan 2026, ss.589-595, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/iceee69936.2026.11598454
- Basıldığı Şehir: Antalya
- Basıldığı Ülke: Türkiye
- Sayfa Sayıları: ss.589-595
- Anahtar Kelimeler: BERTopic, customer feedback analytics, LDA, Topic modeling
- Gazi Üniversitesi Adresli: Evet
Özet
Topic modeling aims to discover latent semantic themes in large collections of documents. Although Latent Dirichlet Allocation (LDA) has long been a standard approach, recent transformer-based methods such as BERTopic use contextual embeddings and class-based TF-IDF to improve topic interpretability. This paper presents an automatic topicmodeling system based on a modified BERTopic pipeline. A dataset consisting of customer reviews was used to evaluate the proposed pipeline, which demonstrates its applicability in realworld scenarios. The proposed pipeline integrates multilingual sentence embeddings, Uniform Manifold Approximation and Projection (UMAP) for dimensionality reduction, Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) for clustering, and class-based TF-IDF (cTF-IDF) for topic representation.