Proposal of Document Classification with Word Sense Disambiguation

Liu, Xiaozhong

Permalink

https://hdl.handle.net/2142/15145 Copy

Description

Title

Proposal of Document Classification with Word Sense Disambiguation

Author(s)

Liu, Xiaozhong

Issue Date

2008-02-28

Keyword(s)

• Document classification
• WSD
• NLP

Abstract

An important area of NLP is the study of Word Sense Disambiguation (WSD), which may assign a unique word sense to a word. There are different methods to implement WSD: one is the word sense based on the collocation of other words (Yarowsky, 1993), where nearby words provide strong consistent clues to the sense of a target word, conditional on relative distance, order and syntactic relationship; and the other is the word sense based on discourse (Gale et al, 1992), where the sense is consistent within any given document. Many experiments in recent years of both supervised (Leacock 1993) and unsupervised (Yarowsky, 1993) WSD algorithms have accomplished promising performance with a high precision rate. Another important area in the field of text mining (Lewis & Spark Jones papers) is document classification, which identifies one or more of several topic labels for a text document. A significant body of research has improved the results of document classification, with innovations in identifying document features as well as improving algorithms. In this study, we will use WSD as part of a method to create innovative features to represent the documents for classification task. With the help of WSD, a set of specially selected ambiguous words can be further distinguished by word sense clustering, in order to achieve better document classification.

Type of Resource

text

Permalink

http://hdl.handle.net/2142/15145

Owning Collections