Premium
Information filtering with extracted index words using ICA
Author(s) -
Yokoi Takeru,
Yanagimoto Hidekazu,
Omatu Sigeru
Publication year - 2009
Publication title -
electronics and communications in japan
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.131
H-Index - 13
eISSN - 1942-9541
pISSN - 1942-9533
DOI - 10.1002/ecj.10018
Subject(s) - dimension (graph theory) , index (typography) , independent component analysis , vector space model , relevance (law) , computer science , artificial intelligence , space (punctuation) , pattern recognition (psychology) , word (group theory) , information retrieval , speech recognition , mathematics , geometry , world wide web , political science , pure mathematics , law , operating system
Abstract We propose an information filtering system with extracted index words using Independent Component Analysis (ICA). Elements of a document vector are established as the weights of index words and their dimensions become larger as the number of documents is increased. Therefore, from the viewpoint of processing time and memory space, the dimension must be decreased. The proposed method decreases the dimension by selecting the index words based on the topics included in the corpus. We have applied ICA to the documents to obtain the topics. Then filtering by the relevance feedback with the document vectors reconstructed by the selected index words, was carried out to confirm the effectiveness of the proposed method. © 2009 Wiley Periodicals, Inc. Electron Comm Jpn, 92(2): 21–27, 2009; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/ecj.10018