Abstract

The project pursued in this paper is to develop from rst information-geometric principles a general method for learning the similarity between text documents. Each individual document is modeled as a memoryless information source. Based on a latent class decomposition of the term-document matrix, a lowdimensional (curved) multinomial subfamily is learned. From this model a canonical similarity function -- known as the Fisher kernel -- is derived. Our approach can be applied for unsupervised and supervised learning problems alike. This in particular covers interesting cases where both, labeled and unlabeled data are available. Experiments in automated indexing and text categorization verify the advantages of the proposed method.

Citations

2168	Indexing by latent semantic analysis - Deerwester, Dumais, et al. - 1990
1368	Text categorization with support vector machines - Joachims - 1998
698	Learning the parts of objects by non-negative matrix factorization - Lee, Seung - 1999
545	Probabilistic latent semantic indexing - Hofmann - 1999
478	L: Distributional Clustering of English Words - Pereira, Tishby, et al. - 1993
315	Exploiting generative models in discriminative classifiers - Jaakkola, Haussler - 1998
89	Unsupervised learning from dyadic data - Hofmann, Puzicha - 1998
63	Aggregate and mixed-order Markov models for statistical language processing - Saul, Pereira - 1997
61	Geometrical Foundations of Asymptotic Inference - Kass, Vos - 1997
41	Stratified Exponential Families: Graphical Models and Model Selection - Geiger, King
25	erential-geometrical methods in statistics - Amari, Di - 1985
14	Latent class analysis of two-way contingency tables by Bayesian methods, Biometrika 76 - Evans, Gilula, et al. - 1989
11	Exponential and Spherical Subfamily Models - Gous - 1998
7	erential Geometry and Statistics - Murray, Rice, et al. - 1993
4	Canonical analysis of contingency tables by maximum likelihood - Gilula, Haberman - 1986