Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/9999
Title: Document ranking for variable bit-block compression signatures
Authors: Luk, RWP 
Chen, CM
Issue Date: 2001
Publisher: Pergamon Press
Source: Information processing and management, 2001, v. 37, no. 1, p. 39-51 How to cite?
Journal: Information processing and management 
Abstract: Variable bit-block compression (VBC) signature is extended for document ranking. Two different extensions were experimented: the weighted VBC (WVBC) scheme and the aggregate VBC (AVBC) scheme. For both, analytical bounds of the additional storage for the term frequencies were derived. The upper and lower bounds of WVBC signatures were better than the corresponding bounds for AVBC signatures. In general, these bounds are functions of the word size (in bits) of the term frequencies. Therefore, term frequencies were scaled to reduce the word size. Experiments showed that the additional storage cost is closer to the lower than the upper bound for both WVBC and AVBC signatures. In addition, WVBC signatures were better than AVBC signatures in terms of storage and retrieval speed. Logarithmic scaling was found to be significantly better than linear scaling, in measuring the agreement of document ranking against the case without scaling, using the Kendall rank-order correlation. If a 75% ranking performance is acceptable, then the additional storage of the term frequencies is only 3.4% of all the indexed documents.
URI: http://hdl.handle.net/10397/9999
ISSN: 0306-4573
DOI: 10.1016/S0306-4573(00)00020-0
Appears in Collections:Journal/Magazine Article

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

SCOPUSTM   
Citations

2
Last Week
0
Last month
0
Citations as of Sep 11, 2017

WEB OF SCIENCETM
Citations

2
Last Week
0
Last month
0
Citations as of Jul 28, 2017

Page view(s)

39
Last Week
2
Last month
Checked on Sep 18, 2017

Google ScholarTM

Check

Altmetric



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