mlsanityData to 7 Oct 2026

CATCH: A Controllable Theme Detection Framework with Contextualized Clustering and Hierarchical Generation

arXiv 2512.21715 ↗First posted 25 Dec 2025cs.CL cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Theme detection is a fundamental task in user-centric dialogue systems, aiming to identify the latent topic of each utterance without relying on predefined schemas. Unlike intent induction, which operates within fixed label spaces, theme detection requires cross-dialogue consistency and alignment with personalized user preferences, posing significant challenges. Existing methods often struggle with sparse, short utterances for accurate topic representation and fail to capture user-level thematic preferences across dialogues. To address these challenges, we propose CATCH (Controllable Theme Detection with Contextualized Clustering and Hierarchical Generation), a unified framework that integrates three core components: (1) context-aware topic representation, which enriches utterance-level semantics using surrounding topic segments; (2) preference-guided topic clustering, which jointly models semantic proximity and personalized feedback to align themes across dialogue; and (3) a hierarchical theme generation mechanism designed to suppress noise and produce robust, coherent topic labels. Experiments on a multi-domain customer dialogue benchmark (DSTC-12) demonstrate the effectiveness of CATCH with 8B LLM in both theme clustering and topic generation quality.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 23 of its 36 references that are in the map. First posted on arXiv 25 Dec 2025.

Builds on

23 of 36 references.

References · 23

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,383
  3. 2022BERTopic: Neural topic modeling with a class-based TF-IDF procedurecited by 779
  4. 2020Tired of Topic Models? Clusters of Pretrained Word Embeddings Make for Fast and Good Topics too!cited by 73
  5. 2023ClusterLLM: Large Language Models as a Guide for Text Clusteringcited by 69
  6. 2022Is Neural Topic Modelling Better than Clustering? An Empirical Study on Clustering with Contextual Embeddings for Topicscited by 35
  7. 2021Improving Unsupervised Dialogue Topic Segmentation with Utterance-Pair Coherence Scoringcited by 29
  8. 2020Intent Mining from past conversations for conversational agentcited by 25
  9. 2023Uncovering the Potential of ChatGPT for Discourse Analysis in Dialogue: An Empirical Studycited by 24
  10. 2019Dialog Intent Induction with Deep Multi-View Clusteringcited by 20
  11. 2021Automatic Intent-Slot Induction for Dialogue Systemscited by 13
  12. 2023Unsupervised Dialogue Topic Segmentation with Topic-aware Utterance Representationcited by 11
11 more
  1. 2023Intent Induction from Conversations for Task-Oriented Dialogue Track at DSTC 11cited by 10
  2. 2021Open Intent Discovery through Unsupervised Semantic Clustering and Dependency Parsingcited by 9
  3. 2024NeuroMax: Enhancing Neural Topic Modeling via Maximizing Mutual Information and Group Topic Regularizationcited by 9
  4. 2020Bag of biterms modeling for short textscited by 4
  5. 2024GloCOM: A Short Text Neural Topic Model via Global Clustering Contextcited by 4
  6. 2020Dynamic transformation of prior knowledge into Bayesian models for data streamscited by 3
  7. 2023Advancing Topic Segmentation and Outline Generation in Chinese Texts: The Paragraph-level Topic Representation, Corpus, and Benchmarkcited by 3
  8. 2024Unsupervised Mutual Learning of Discourse Parsing and Topic Segmentation in Dialoguecited by 3
  9. 2022Dialog Intent Induction via Density-based Deep Clustering Ensemblecited by 2
  10. 2023Towards a Fully Unsupervised Framework for Intent Induction in Customer Support Dialoguescited by 1
  11. 2025Overview of Dialog System Evaluation Track: Dimensionality, Language, Culture and Safety at DSTC 12cited by 1

Built on by

Not cited yet.