Structural Poisson Mixtures for Classification of Documents

0317515 - ÚTIA 2010 RIV US eng C - Conference Paper (international conference)
Grim, Jiří - Novovičová, Jana - Somol, Petr
Structural Poisson Mixtures for Classification of Documents.
[Strukturní Poissonovské směsi pro klasifikaci dokumentů.]
Proceedings of the 19th International Conference on Pattern Recognition. Los Alamitos: IEEE Press, 2008, s. 1324-1327. ISBN 978-1-4244-2174-9.
[19th International Conference on Pattern Recognition. Tampa (US), 07.12.2008-11.12.2008]
R&D Projects: GA MŠMT 1M0572; GA ČR GA102/07/1594
Grant - others:GA MŠk(CZ) 2C06019
Institutional research plan: CEZ:AV0Z10750506
Keywords : classification of documents * Poisson mixtures * Structural approach
Subject RIV: IN - Informatics, Computer Science
http://library.utia.cas.cz/separaty/2008/RO/grim-structural poisson mixtures for classification of documents.pdf

Considering the statistical text classification problem we approximate class-conditional probability distributions by structurally modified Poisson mixtures. By introducing the structural model we can use different subsets of input variables to evaluate conditional probabilities of different classes in the Bayes formula. The method is applicable to document vectors of arbitrary dimension without any preprocessing. The structural optimization can be included into the EM algorithm in a statistically correct way.

V rámci statistického přístupu k problému klasifikace dokumentů jsou dokumenty reprezentovány formou /bag-of-words/. Podmíněné distribuce dokumentů v jednotlivých třídách jsou aproximovány ve tvaru strukturní poissonovské distribuční směsi. Bayesovská klasifikace dokumentů je ověřována na datových souborech Reuters a 20 NEWSGROUPS.
Permanent Link: http://hdl.handle.net/11104/0167137