Iterative subspace text categorization

Pao Yue-kong Library Electronic Theses Database

Iterative subspace text categorization

 

Author: Chik, Cho-yiu Francis
Title: Iterative subspace text categorization
Degree: M.Phil.
Year: 2013
Subject: Text processing (Computer science)
Artificial intelligence.
Hong Kong Polytechnic University -- Dissertations
Department: Dept. of Computing
Pages: xi, 152 leaves : ill. ; 30 cm.
InnoPac Record: http://library.polyu.edu.hk/record=b2652711
URI: http://theses.lib.polyu.edu.hk/handle/200/7229
Abstract: Text categorization finds many practical applications. The dominant approach involves the use of various machine learning techniques where classification rules are automatically created using information from labeled texts. The proposed method to combat the curse of dimensionality is subspace methodology. However, this has only been applied broadly in unsupervised text categorization. The performance of subspace methodology on supervised text categorization has not yet been found. The approach of iterative subspace method of pattern classification is investigated. For the topic pairs of "carcass_livestock" and "soybean_oilseed" from the Reuters-21578 collection, the results with confidence level greater than 95% under 8-fold/10-fold/12-fold cross validation shows the potential of this approach. It is expected that the performance can be further improved by using other optimization techniques. It is still promising that there is 8.24% precision improvement of "livestock" evaluated comparing to 1-level classifier, standard Support Vector Machine (SVM), under 8-fold cross validation. There is also 11.85% improvement of "nat-gas" evaluated comparing to Soft Margin SVM classifier under 8-fold cross validation.

Files in this item

Files Size Format
b2652711x.pdf 1.681Mb PDF
Copyright Undertaking
As a bona fide Library user, I declare that:
  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

     

Quick Search

Browse

More Information