MULTI-LEVEL TEXT CLASSIFICATION METHOD BASED ON LATENT SEMANTIC ANALYSIS

Hongxia Shi, Guiyi Wei, Yun Pan

2007

Abstract

In multi-level text classification, all categories have level relation. The categories in the same layer have a certain generality. By applying LSA theory to multi-level text classification, the words’ semantic relationship is better represented and their weight equations are adjusted so they are more reasonable. This method extends the traditional vector space model to LSA space model and consequent experiments got very good results.

References

  1. Liu Q, Li S J. The calculation of semantic similarity between vocabularies based on “http://www.keenage.com”. http://www.nlp.org.cn/categories/default.php?cat_id=1 4
  2. Zhou S G, Guan J H, Hu Y F. Latent semantic indexing (LSI) and its applications in applications in Chinese text processing [J]. Mini-Micro Systems, (2001), (2): 987-991
  3. Huang H Y, Lin S M, Yan X W. A study of text classification based on concept space [J] Computer Science, (2003), 30(3): 46-49
  4. Wang G Y, Xu J S. A new method of text categorization based on LSA and Kohonen network [J] Computer Applications, (2004), 24(2)
  5. Schapire R, Singer Y.Boos Texter:a boosting-based system for text categorization . Machine Learning(2000), 39(2/3): 135-168
  6. Campbell C,Cristianin N,Smo1a A.Query Learning with Large Margin Classifiers[A] . proceedings of the Seventeenth International Conference on Machine Learning[C].(2000).111-118
  7. Hsu C W,Lin C J.A Compare of Methods for Multi-class Support Vector Machine [Z], (2001)
Download


Paper Citation


in Harvard Style

Shi H., Wei G. and Pan Y. (2007). MULTI-LEVEL TEXT CLASSIFICATION METHOD BASED ON LATENT SEMANTIC ANALYSIS . In Proceedings of the Ninth International Conference on Enterprise Information Systems - Volume 4: ICEIS, ISBN 978-972-8865-91-7, pages 320-324. DOI: 10.5220/0002401703200324


in Bibtex Style

@conference{iceis07,
author={Hongxia Shi and Guiyi Wei and Yun Pan},
title={MULTI-LEVEL TEXT CLASSIFICATION METHOD BASED ON LATENT SEMANTIC ANALYSIS},
booktitle={Proceedings of the Ninth International Conference on Enterprise Information Systems - Volume 4: ICEIS,},
year={2007},
pages={320-324},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0002401703200324},
isbn={978-972-8865-91-7},
}


in EndNote Style

TY - CONF
JO - Proceedings of the Ninth International Conference on Enterprise Information Systems - Volume 4: ICEIS,
TI - MULTI-LEVEL TEXT CLASSIFICATION METHOD BASED ON LATENT SEMANTIC ANALYSIS
SN - 978-972-8865-91-7
AU - Shi H.
AU - Wei G.
AU - Pan Y.
PY - 2007
SP - 320
EP - 324
DO - 10.5220/0002401703200324