Title :
A new text dimension reduction method based on PA and LSI
Author :
Wang, Ming-chun ; Xu, Jian-Suo ; Wang, Zheng-Ou
Author_Institution :
Inst. of Syst. Eng., Tianjin Univ., China
Abstract :
This paper presents a new dimension reduction method based on pattern aggregation theory and latent semantic indexing theory. Our method firstly reduces text dimension with PA method that not only uses class label but also reduces quantities of calculation in the next step, and then makes the dimension much lower by LSI method. Experiments demonstrate the effects of the method.
Keywords :
indexing; information retrieval; pattern classification; text analysis; information retrieval; latent semantic indexing theory; pattern aggregation theory; pattern classification; text dimension reduction method; Algebra; Indexing; Information retrieval; Large scale integration; Matrix decomposition; Statistics; Systems engineering and theory; Text categorization; Text mining; Vocabulary;
Conference_Titel :
Machine Learning and Cybernetics, 2004. Proceedings of 2004 International Conference on
Print_ISBN :
0-7803-8403-2
DOI :
10.1109/ICMLC.2004.1381996
