Information Retrieval

, Volume 8, Issue 2, pp 181–196 | Cite as

Test Data Likelihood for PLSA Models

  • Thorsten BrantsEmail author


Probabilistic Latent Semantic Analysis (PLSA) is a statistical latent class model that has recently received considerable attention. In its usual formulation it cannot assign likelihoods to unseen documents. Furthermore, it assigns a probability of zero to unseen documents during training. We point out that one of the two existing alternative formulations of the Expectation-Maximization algorithms for PLSA does not require this assumption. However, even that formulation does not allow calculation ofthe actual likelihood values. We therefore derive a new test-data likelihood substitute for PLSA and compare it to three existing likelihood substitutes. An empirical evaluation shows that our new likelihood substitute produces the best predictions about accuracies in two different IR tasks and is therefore best suited to determine the number of EM steps when training PLSA models. The new likelihood measure and its evaluation also suggest that PLSA is not very sensitive to overfitting for the two tasks considered. This renders additions like tempered EM that especially address overfitting unnecessary.


Probabilistic Latent Semantic Analysis PLSA likelihood 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. Beeferman D, Berger A and Lafferty J (1999) Statistical models for text segmentation. Machine Learning 34(1–3):177–210.Google Scholar
  2. Blei D, Ng A and Jordan M (2001) Latent dirichlet allocation. In: Proceedings of NIPS-2001, Vancouver, BC, Canada, pp. 601–608.Google Scholar
  3. Brants T, Chen F and Tsochantaridis I (2002) Topic-based document segmentation with probabilistic latent semantic analysis. In: International Conference on Information and Knowledge Management (CIKM), McLean, VA, pp. 211–218.Google Scholar
  4. Choi FYY, Wiemer-Hastings P and Moore J (2001) Latent semantic analysis for text segmentation. In: Lee L and Harman D, Eds., In Proceedings of the 2001 Conference on Empirical Methods in Natural Language Processing, pp. 109–117.Google Scholar
  5. Dempster AP, Laird NM and Rubin DB (1977) Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society, 39(1):1–21.Google Scholar
  6. Gildea D and Hofmann T (1999) Topic based language models using EM. In: Proceedings of 6th European Conference On Speech Communication and Technology (Eurospeech’99), Budapest, Hungary, pp. 2167–2170.Google Scholar
  7. Hearst MA (1997) TextTiling: Segmenting text into multi-paragraph subtopic passages. Computational Linguistics, 23(1):33–64.Google Scholar
  8. Hofmann T (1999) Probabilistic latent semantic analysis. In: Proceedings of Uncertainty in Artificial Intelligence, Stockholm, Sweden, pp. 289–296.Google Scholar
  9. Hofmann T (2000) Probabilistic latent semantic indexing. In: Proceedings of SIGIR-99, Berkeley, CA, pp. 35–44.Google Scholar
  10. Hofmann T (2001) Unsupervised learning by probabilistic latent semantic analysis. Machine Learning, 42:177–196.Google Scholar
  11. Hofmann T and Puzicha J (1998) Unsupervised learning from dyadic data. Technical Report TR-98-042, ICSI, Berkeley, CA.Google Scholar
  12. Li H and Yamanishi K (2000) Topic analysis using a finite mixture model. In: Proceedings of Joint SIGDAT Conference on Empirical Methods in Natural Language Processing and Very Large Corpora, pp. 35–44.Google Scholar
  13. Rooth M, Riezler S, Prescher D, Carroll G, and Beil F (1999) Inducing a semantically annotated lexicon via EM-based clustering. In: Proceedings of ACL-99, College Park, MD, USA, pp. 104–111.Google Scholar
  14. Saul L and Pereira F (1997) Aggregate and mixed-order markov models for statistical language processing. In: Proceedings of the Second Conference on Empirical Methods in Natural Language Processing (EMNLP), San Francisco, CA, Association for Computational Linguistics, pp. 81–89.Google Scholar

Copyright information

© Springer Science + Business Media, Inc. 2005

Authors and Affiliations

  1. 1.Google, Inc.Mountain ViewUSA

Personalised recommendations