Comparing relevance feedback algorithms for web search

Vishwa Vinay; Ken Wood; Natasa Milic-Frayling; Ingemar J. Cox

doi:10.1145/1062745.1062864

Comparing relevance feedback algorithms for web search

Vishwa Vinay^*, Ken Wood, Natasa Milic-Frayling, Ingemar J. Cox

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

14 Citations (Scopus)

Abstract

We evaluate three different relevance feedback (RF)algorithms, Rocchio, Robertson/Sparck-Jones (RSJ)and Bayesian, in the context of Web search. We use a target-testing experimental procedure whereby a user must locate a specific document. For user relevance feedback, we consider all possible user choices of indicating zero or more relevant documents from a set of 10 displayed documents. Examination of the effects of each user choice permits us to compute an upper-bound on the performance of each RF algorithm.We ind that there is a significant variation in the upper-bound performance o the three RF algorithms and that the Bayesian algorithm approaches the best possible.

Original language	English
Title of host publication	14th International World Wide Web Conference, WWW2005
Pages	1052-1053
Number of pages	2
DOIs	https://doi.org/10.1145/1062745.1062864
Publication status	Published - 2005
Externally published	Yes
Event	14th International World Wide Web Conference, WWW2005 - Chiba, Japan Duration: 10 May 2005 → 14 May 2005

Publication series

Name	14th International World Wide Web Conference, WWW2005

Conference

Conference	14th International World Wide Web Conference, WWW2005
Country/Territory	Japan
City	Chiba
Period	10/05/05 → 14/05/05

Keywords

Evaluation
Relevance feedback
Web search

Access to Document

10.1145/1062745.1062864

Cite this

@inproceedings{a8c1677fdc08460d80abbe42acc84d53,

title = "Comparing relevance feedback algorithms for web search",

abstract = "We evaluate three different relevance feedback (RF)algorithms, Rocchio, Robertson/Sparck-Jones (RSJ)and Bayesian, in the context of Web search. We use a target-testing experimental procedure whereby a user must locate a specific document. For user relevance feedback, we consider all possible user choices of indicating zero or more relevant documents from a set of 10 displayed documents. Examination of the effects of each user choice permits us to compute an upper-bound on the performance of each RF algorithm.We ind that there is a significant variation in the upper-bound performance o the three RF algorithms and that the Bayesian algorithm approaches the best possible.",

keywords = "Evaluation, Relevance feedback, Web search",

author = "Vishwa Vinay and Ken Wood and Natasa Milic-Frayling and Cox, {Ingemar J.}",

year = "2005",

doi = "10.1145/1062745.1062864",

language = "English",

isbn = "1595930515",

series = "14th International World Wide Web Conference, WWW2005",

pages = "1052--1053",

booktitle = "14th International World Wide Web Conference, WWW2005",

note = "14th International World Wide Web Conference, WWW2005 ; Conference date: 10-05-2005 Through 14-05-2005",

}

TY - GEN

T1 - Comparing relevance feedback algorithms for web search

AU - Vinay, Vishwa

AU - Wood, Ken

AU - Milic-Frayling, Natasa

AU - Cox, Ingemar J.

PY - 2005

Y1 - 2005

N2 - We evaluate three different relevance feedback (RF)algorithms, Rocchio, Robertson/Sparck-Jones (RSJ)and Bayesian, in the context of Web search. We use a target-testing experimental procedure whereby a user must locate a specific document. For user relevance feedback, we consider all possible user choices of indicating zero or more relevant documents from a set of 10 displayed documents. Examination of the effects of each user choice permits us to compute an upper-bound on the performance of each RF algorithm.We ind that there is a significant variation in the upper-bound performance o the three RF algorithms and that the Bayesian algorithm approaches the best possible.

AB - We evaluate three different relevance feedback (RF)algorithms, Rocchio, Robertson/Sparck-Jones (RSJ)and Bayesian, in the context of Web search. We use a target-testing experimental procedure whereby a user must locate a specific document. For user relevance feedback, we consider all possible user choices of indicating zero or more relevant documents from a set of 10 displayed documents. Examination of the effects of each user choice permits us to compute an upper-bound on the performance of each RF algorithm.We ind that there is a significant variation in the upper-bound performance o the three RF algorithms and that the Bayesian algorithm approaches the best possible.

KW - Evaluation

KW - Relevance feedback

KW - Web search

UR - http://www.scopus.com/inward/record.url?scp=77953049528&partnerID=8YFLogxK

U2 - 10.1145/1062745.1062864

DO - 10.1145/1062745.1062864

M3 - Conference contribution

AN - SCOPUS:77953049528

SN - 1595930515

SN - 9781595930514

T3 - 14th International World Wide Web Conference, WWW2005

SP - 1052

EP - 1053

BT - 14th International World Wide Web Conference, WWW2005

T2 - 14th International World Wide Web Conference, WWW2005

Y2 - 10 May 2005 through 14 May 2005

ER -

Comparing relevance feedback algorithms for web search

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this