TriTopic: Tri-Modal Graph-Based Topic Modeling with Iterative Refinement and Archetypes
This work addresses critical issues in topic modeling for researchers and practitioners, offering a more stable and precise method, though it appears incremental as it builds on existing approaches like BERTopic.
The paper tackles limitations in topic modeling such as instability and loss of lexical precision by introducing TriTopic, a tri-modal graph-based framework that achieves the highest NMI scores across multiple datasets, with a mean NMI of 0.575 compared to 0.513 for BERTopic.
Topic modeling extracts latent themes from large text collections, but leading approaches like BERTopic face critical limitations: stochastic instability, loss of lexical precision ("Embedding Blur"), and reliance on a single data perspective. We present TriTopic, a framework that addresses these weaknesses through a tri-modal graph fusing semantic embeddings, TF-IDF, and metadata. Three core innovations drive its performance: hybrid graph construction via Mutual kNN and Shared Nearest Neighbors to eliminate noise and combat the curse of dimensionality; Consensus Leiden Clustering for reproducible, stable partitions; and Iterative Refinement that sharpens embeddings through dynamic centroid-pulling. TriTopic also replaces the "average document" concept with archetype-based topic representations defined by boundary cases rather than centers alone. In benchmarks across 20 Newsgroups, BBC News, AG News, and Arxiv, TriTopic achieves the highest NMI on every dataset (mean NMI 0.575 vs. 0.513 for BERTopic, 0.416 for NMF, 0.299 for LDA), guarantees 100% corpus coverage with 0% outliers, and is available as an open-source PyPI library.