Individual Difference, Query Reformulation and Search Performance in Ad-hoc Search Task

Ying-Hsang Liu

Research output: Book chapter/Published conference paperConference paperpeer-review

2 Citations (Scopus)
8 Downloads (Pure)


This paper reports on findings from a larger study that was designed to assess the effectiveness of MeSH (Medical Subject Headings) terms when used by different types of searchers in an interactive search environment. Specifically, the impact of individual differences of domain knowledge and search expertise on querying behavior and user search performance is assessed, using the search logs from an interactive information retrieval experiment. The results show that there are significant differences in query reformulations characterized by the time spent and number of query iterations. From both single query and session-based perspectives of evaluation strategy in evaluating queries over search sessions, it establishes the correlation between the individual difference and the user search performance. More importantly, search novices have not benefited from query reformulations in searching relatively difficult search topics for ad-hoc search tasks in terms of the precision measure. Implications for interactive information retrieval system evaluation are discussed.
Original languageEnglish
Title of host publicationProceedings of the Lernen, Wissen, Adaptivität (LWA) Symposium 2011
Subtitle of host publicationLearning, Knowledge, Adaption
EditorsMyra Spiliopoulou, Andreas Nurnberger, Rene Schult
Place of PublicationGermany
PublisherUniversitat Magdeburg
Number of pages8
Publication statusPublished - 2011
Event Lernen, Wissen, Adaptivität (LWA) Symposium 2011 - Fakultät für Informatik, Otto-von-Guericke-Universität, Magdeburg, Germany
Duration: 28 Sep 201130 Sep 2011 (Conference TOC)


Workshop Lernen, Wissen, Adaptivität (LWA) Symposium 2011
Abbreviated titleLearning, Knowledge, Adaption
Internet address

Fingerprint Dive into the research topics of 'Individual Difference, Query Reformulation and Search Performance in Ad-hoc Search Task'. Together they form a unique fingerprint.

Cite this