Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/70235
Title: Topic sequence kernel
Authors: Xu, J
Lu, Q 
Liu, Z
Keywords: Topic sequence
String kernel
Classification
Issue Date: 2012
Publisher: Springer
Source: Lecture notes in computer science (including subseries Lecture notes in artificial intelligence and lecture notes in bioinformatics), 2012, v. 7675, p. 457-466 How to cite?
Journal: Lecture notes in computer science (including subseries Lecture notes in artificial intelligence and lecture notes in bioinformatics) 
Abstract: This paper addresses the problem of classifying documents using the kernel approaches based on topic sequences. Previously, the string kernel uses the ordered subsequence of characters as features and the word sequence kernel is proposed to use words as the subsequences. However, they both face the problem of computational complexity because of the large amount of symbols (characters or words). This paper, therefore, proposes to use sequences of topics rather than characters or words to reduce the number of symbols, thus increasing the computational efficiency. Documents that exhibit similar posterior topic proportions are expected to have similar topic sequence and then should be classified into the same category. Experiments conducted on the Reuters-21578 datasets have proven this hypothesis.
Description: 8th Asia Information Retrieval Societies Conference, AIRS 2012, Tianjin, China, December 17-19, 2012
URI: http://hdl.handle.net/10397/70235
ISBN: 978-3-642-35340-6
978-3-642-35341-3
ISSN: 0302-9743
EISSN: 1611-3349
DOI: 10.1007/978-3-642-35341-3_41
Appears in Collections:Conference Paper

Access
View full-text via PolyU eLinks SFX Query
Show full item record

SCOPUSTM   
Citations

1
Last Week
0
Last month
Citations as of May 31, 2018

Page view(s)

48
Last Week
3
Last month
Citations as of Jun 18, 2018

Google ScholarTM

Check

Altmetric


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.