Overview

Topic modeling finds recurring themes in text and shows how strongly each document relates to them, helping researchers explore large collections such as scientific papers, surveys, and social media posts. Approaches range from probabilistic models such as LDA and Structural Topic Models to embedding-based clustering, neural and graph models, and systems that use large language models to refine or explain topics. Researchers evaluate whether topics are coherent and distinct, whether document assignments make sense, and whether the results support tasks such as classification or tracking change over time. A central challenge is balancing semantic richness with interpretability, stability, and comparability, since automated metrics do not always match human judgments and performance can vary across domains and text lengths.

Latest papers

CardsList

62 papers6 papers in the last four weeks, up 50% on the four weeks before. 0.0% of all new papers.

  1. FLAME: A New Dataset on FLemish Accounts of Momentary Experiences

    Apr 20, 2025Ratna Kandala, Niels Vanhasbroeck, Katie HoemannTopic ModelingLow-Resource Language Processing

  2. Research on Domain Information Mining and Theme Evolution of Scientific Papers

    Apr 18, 2022Changwei Zheng, Zhe Xue, Meiyu Liang +2Representation LearningTopic Modeling