Stable topic modeling with local density regularization

Sergei Koltcov, Sergey I. Nikolenko, Olessia Koltsova, Vladimir Filippov, Svetlana Bodrunova

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

12 Scopus citations


Topic modeling has emerged over the last decade as a powerful tool for analyzing large text corpora, including Web-based usergenerated texts. Topic stability, however, remains a concern: topic models have a very complex optimization landscape with many local maxima, and even different runs of the same model yield very different topics. Aiming to add stability to topic modeling, we propose an approach to topic modeling based on local density regularization, where words in a local context window of a given word have higher probabilities to get the same topic as that word. We compare several models with local density regularizers and show how they can improve topic stability while remaining on par with classical models in terms of quality metrics.

Original languageEnglish
Title of host publicationInternet Science - 3rd International Conference, INSCI 2016, Proceedings
EditorsAnna Satsiou, Yanina Welp, Thanassis Tiropanis, Dominic DiFranzo, Ioannis Stavrakakis, Franco Bagnoli, Paolo Nesi, Giovanna Pacini
PublisherSpringer Nature
Number of pages13
ISBN (Print)9783319459813
StatePublished - 2016
Event3rd International Conference on Internet Science, INSCI 2016 - Florence, Italy
Duration: 12 Sep 201614 Sep 2016

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9934 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference3rd International Conference on Internet Science, INSCI 2016

Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)


  • Gibbs sampling
  • Latent Dirichlet allocation
  • Topic modeling

Fingerprint Dive into the research topics of 'Stable topic modeling with local density regularization'. Together they form a unique fingerprint.

Cite this