{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/parsimonious-topic-models-with-salient-word","title":"Parsimonious Topic Models with Salient Word Discovery","arxiv_id":"1401.6169","date":"2014-01-22","proceeding":null,"authors":["Hossein Soleimani","David J. Miller"],"abstract":"We propose a parsimonious topic model for text corpora. In related models\nsuch as Latent Dirichlet Allocation (LDA), all words are modeled\ntopic-specifically, even though many words occur with similar frequencies\nacross different topics. Our modeling determines salient words for each topic,\nwhich have topic-specific probabilities, with the rest explained by a universal\nshared model. Further, in LDA all topics are in principle present in every\ndocument. By contrast our model gives sparse topic representation, determining\nthe (small) subset of relevant topics for each document. We derive a Bayesian\nInformation Criterion (BIC), balancing model complexity and goodness of fit.\nHere, interestingly, we identify an effective sample size and corresponding\npenalty specific to each parameter type in our model. We minimize BIC to\njointly determine our entire model -- the topic-specific words,\ndocument-specific topics, all model parameter values, {\\it and} the total\nnumber of topics -- in a wholly unsupervised fashion. Results on three text\ncorpora and an image dataset show that our model achieves higher test set\nlikelihood and better agreement with ground-truth class labels, compared to LDA\nand to a model designed to incorporate sparsity.","url_abs":"http://arxiv.org/abs/1401.6169v2","url_pdf":"http://arxiv.org/pdf/1401.6169v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"parsimonious-topic-models-with-salient-word","repo_url":"https://github.com/hsoleimani/PTM","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"topic-models","task_name":"Topic Models"}],"methods":[{"method_slug":"lda","method_name":"LDA"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}