An analysis of failed queries for web image retrieval

Research output: Contribution to journalArticle

28 Citations (Scopus)

Abstract

This paper examines a large number of failed queries submitted to a web image search engine, including real users' search terms and written requests. The results show that failed image queries have a much higher specificity than successful queries because users often employ various refined types to specify their queries. The study explores the refined types further, and finds that failed queries consist of far more conceptual than perceptual refined types. The widely used content-based image retrieval technique, CBIR, can only deal with a small proportion of failed queries; hence, appropriate integration of concept-based techniques is desirable. Based on using the concepts of uniqueness and refinement for categorization, the study also provides a useful discussion on the gaps between image queries and retrieval techniques. The initial results enhance the understanding of failed queries and suggest possible ways to improve image retrieval systems.

Original languageEnglish
Pages (from-to)275-289
Number of pages15
JournalJournal of Information Science
Volume34
Issue number3
DOIs
Publication statusPublished - 2008 Jun 1

Fingerprint

Image retrieval
Search engines
World Wide Web
search engine

Keywords

  • Image retrieval
  • Image user studies
  • Query log analysis
  • Web image search engines

ASJC Scopus subject areas

  • Information Systems
  • Library and Information Sciences

Cite this

An analysis of failed queries for web image retrieval. / Pu, Hsiao Tieh.

In: Journal of Information Science, Vol. 34, No. 3, 01.06.2008, p. 275-289.

Research output: Contribution to journalArticle

@article{2f55224faad9480180b559d7b7eaa267,
title = "An analysis of failed queries for web image retrieval",
abstract = "This paper examines a large number of failed queries submitted to a web image search engine, including real users' search terms and written requests. The results show that failed image queries have a much higher specificity than successful queries because users often employ various refined types to specify their queries. The study explores the refined types further, and finds that failed queries consist of far more conceptual than perceptual refined types. The widely used content-based image retrieval technique, CBIR, can only deal with a small proportion of failed queries; hence, appropriate integration of concept-based techniques is desirable. Based on using the concepts of uniqueness and refinement for categorization, the study also provides a useful discussion on the gaps between image queries and retrieval techniques. The initial results enhance the understanding of failed queries and suggest possible ways to improve image retrieval systems.",
keywords = "Image retrieval, Image user studies, Query log analysis, Web image search engines",
author = "Pu, {Hsiao Tieh}",
year = "2008",
month = "6",
day = "1",
doi = "10.1177/0165551507084140",
language = "English",
volume = "34",
pages = "275--289",
journal = "Journal of Information Science",
issn = "0165-5515",
publisher = "SAGE Publications Ltd",
number = "3",

}

TY - JOUR

T1 - An analysis of failed queries for web image retrieval

AU - Pu, Hsiao Tieh

PY - 2008/6/1

Y1 - 2008/6/1

N2 - This paper examines a large number of failed queries submitted to a web image search engine, including real users' search terms and written requests. The results show that failed image queries have a much higher specificity than successful queries because users often employ various refined types to specify their queries. The study explores the refined types further, and finds that failed queries consist of far more conceptual than perceptual refined types. The widely used content-based image retrieval technique, CBIR, can only deal with a small proportion of failed queries; hence, appropriate integration of concept-based techniques is desirable. Based on using the concepts of uniqueness and refinement for categorization, the study also provides a useful discussion on the gaps between image queries and retrieval techniques. The initial results enhance the understanding of failed queries and suggest possible ways to improve image retrieval systems.

AB - This paper examines a large number of failed queries submitted to a web image search engine, including real users' search terms and written requests. The results show that failed image queries have a much higher specificity than successful queries because users often employ various refined types to specify their queries. The study explores the refined types further, and finds that failed queries consist of far more conceptual than perceptual refined types. The widely used content-based image retrieval technique, CBIR, can only deal with a small proportion of failed queries; hence, appropriate integration of concept-based techniques is desirable. Based on using the concepts of uniqueness and refinement for categorization, the study also provides a useful discussion on the gaps between image queries and retrieval techniques. The initial results enhance the understanding of failed queries and suggest possible ways to improve image retrieval systems.

KW - Image retrieval

KW - Image user studies

KW - Query log analysis

KW - Web image search engines

UR - http://www.scopus.com/inward/record.url?scp=43449135346&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=43449135346&partnerID=8YFLogxK

U2 - 10.1177/0165551507084140

DO - 10.1177/0165551507084140

M3 - Article

AN - SCOPUS:43449135346

VL - 34

SP - 275

EP - 289

JO - Journal of Information Science

JF - Journal of Information Science

SN - 0165-5515

IS - 3

ER -