Improving Neural Topic Modeling with Semantically-Grounded Soft Label Distributions

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Raymond, Abaskohi, Amirhossein, Li, Chuyuan, Murray, Gabriel, Carenini, Giuseppe
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912915145097216
author Li, Raymond
Abaskohi, Amirhossein
Li, Chuyuan
Murray, Gabriel
Carenini, Giuseppe
author_facet Li, Raymond
Abaskohi, Amirhossein
Li, Chuyuan
Murray, Gabriel
Carenini, Giuseppe
contents Traditional neural topic models are typically optimized by reconstructing the document's Bag-of-Words (BoW) representations, overlooking contextual information and struggling with data sparsity. In this work, we propose a novel approach to construct semantically-grounded soft label targets using Language Models (LMs) by projecting the next token probabilities, conditioned on a specialized prompt, onto a pre-defined vocabulary to obtain contextually enriched supervision signals. By training the topic models to reconstruct the soft labels using the LM hidden states, our method produces higher-quality topics that are more closely aligned with the underlying thematic structure of the corpus. Experiments on three datasets show that our method achieves substantial improvements in topic coherence, purity over existing baselines. Additionally, we also introduce a retrieval-based metric, which shows that our approach significantly outperforms existing methods in identifying semantically similar documents, highlighting its effectiveness for retrieval-oriented applications.
format Preprint
id arxiv_https___arxiv_org_abs_2602_17907
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Improving Neural Topic Modeling with Semantically-Grounded Soft Label Distributions
Li, Raymond
Abaskohi, Amirhossein
Li, Chuyuan
Murray, Gabriel
Carenini, Giuseppe
Computation and Language
Artificial Intelligence
Traditional neural topic models are typically optimized by reconstructing the document's Bag-of-Words (BoW) representations, overlooking contextual information and struggling with data sparsity. In this work, we propose a novel approach to construct semantically-grounded soft label targets using Language Models (LMs) by projecting the next token probabilities, conditioned on a specialized prompt, onto a pre-defined vocabulary to obtain contextually enriched supervision signals. By training the topic models to reconstruct the soft labels using the LM hidden states, our method produces higher-quality topics that are more closely aligned with the underlying thematic structure of the corpus. Experiments on three datasets show that our method achieves substantial improvements in topic coherence, purity over existing baselines. Additionally, we also introduce a retrieval-based metric, which shows that our approach significantly outperforms existing methods in identifying semantically similar documents, highlighting its effectiveness for retrieval-oriented applications.
title Improving Neural Topic Modeling with Semantically-Grounded Soft Label Distributions
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2602.17907