%0 Journal Article %1 citeulike:378385 %A Hofmann, Thomas %C Hingham, MA, USA %D 2001 %I Kluwer Academic Publishers %J Machine Learning %K mustread topicinference socialnets %N 1-2 %P 177--196 %T Unsupervised Learning by Probabilistic Latent Semantic Analysis %U http://portal.acm.org/citation.cfm?id=599609.599631 %V 42

@article{citeulike:378385, added-at = {2006-06-16T10:34:37.000+0200}, address = {Hingham, MA, USA}, author = {Hofmann, Thomas}, biburl = {https://www.bibsonomy.org/bibtex/2e7ee4a5fe0f2dd77ddfe5f47d3b9272c/ldietz}, citeulike-article-id = {378385}, comment = {laut Gregor ein besseres einsteiger paper als LDA --- probabilistic version of LSA (which is originally based on Matrix decomposition) introduces latent variables, that can be interpreted as topics. Via which documents can be describes as mixtures of topics (so called aspects) which in turn stand for a word distribution. In contrast to probabilistic clustering algorithms, which try to assign a document to a single cluster, this model aims at assigning a document to a mix of aspects. This model has a probabilistic, a geometric and a matrix interpretation. Domains: Information retrieval (LSI), noun-adjective detection, and others. Task: infer topics of a given (known or unknown) document. match queries against documents. explanation of similarities between documents. Method: Bayesian Model, EM with temperature strategy (although they state, that a poor local maximum is a better than a solution of LSA) Problems: overfitting, [generalization -> stated by Blei's LDA paper]}, interhash = {4c71aa69f5b098f6ac8e1df59959e684}, intrahash = {e7ee4a5fe0f2dd77ddfe5f47d3b9272c}, issn = {0885-6125}, journal = {Machine Learning}, keywords = {mustread topicinference socialnets}, number = {1-2}, pages = {177--196}, priority = {0}, publisher = {Kluwer Academic Publishers}, timestamp = {2006-06-16T10:34:37.000+0200}, title = {Unsupervised Learning by Probabilistic Latent Semantic Analysis}, url = {http://portal.acm.org/citation.cfm?id=599609.599631}, volume = 42, year = 2001 }