Combining Thesaurus Knowledge and Probabilistic Topic Models

TL;DR

This paper introduces a method to enhance probabilistic topic models by integrating thesaurus knowledge, which boosts the contribution of semantically related words in topic discovery, especially when using domain-specific thesauri.

Contribution

The paper presents a novel approach to incorporate thesaurus knowledge into probabilistic topic models, improving their ability to capture semantic relations in texts.

Findings

Using domain-specific thesauri improves topic model quality.

Excluding hyponymy relations with WordNet enhances model performance.

Thesaurus integration benefits semantic coherence in topics.

Abstract

In this paper we present the approach of introducing thesaurus knowledge into probabilistic topic models. The main idea of the approach is based on the assumption that the frequencies of semantically related words and phrases, which are met in the same texts, should be enhanced: this action leads to their larger contribution into topics found in these texts. We have conducted experiments with several thesauri and found that for improving topic models, it is useful to utilize domain-specific knowledge. If a general thesaurus, such as WordNet, is used, the thesaurus-based improvement of topic models can be achieved with excluding hyponymy relations in combined topic models.