Presentation 2003/10/30
Modeling Category Structures with a Kernel Function(Natural Language Understanding and Models of Communication)
Hiroya Takamura, Yuji Matsumoto, Hiroyasu Yamada,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We propose one type of TOP (Tangent vector Of the Posterior log-odds) kernels and apply itto text categorization. In a number of categorization tasks including text categorization, negativeexamples are usually more common than positive examples and there may be several differenttypes of negative examples. Therefore, we regard the probabilistic model of negative examples as a mixture of several models respectively corresponding to given categories. Since each model of ours is expressed using a one-dimensional Gaussian-type function, our kernel has an advantage in computational time. In our experiments, our kernel, combined with SVMs, outperformed the linear kernel and the Fisher kernel based on the Probabilistic Latent Semantic Indexing model.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) text categorization / TOP kernel
Paper # NLC2003-32
Date of Issue

Conference Information
Committee NLC
Conference Date 2003/10/30(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Natural Language Understanding and Models of Communication (NLC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Modeling Category Structures with a Kernel Function(Natural Language Understanding and Models of Communication)
Sub Title (in English)
Keyword(1) text categorization
Keyword(2) TOP kernel
1st Author's Name Hiroya Takamura
1st Author's Affiliation Precision and Intelligence Laboratory, Tokyo Institute of Technology()
2nd Author's Name Yuji Matsumoto
2nd Author's Affiliation Graduate School of Information Science, Nara Institute of Science and Technology
3rd Author's Name Hiroyasu Yamada
3rd Author's Affiliation Graduate School of Information Science. Japan Institute of Science and Technology
Date 2003/10/30
Paper # NLC2003-32
Volume (vol) vol.103
Number (no) 407
Page pp.pp.-
#Pages 8
Date of Issue