Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Ephemeral relevance and user activities in a search session

Jiang, Jiepu (2017) Ephemeral relevance and user activities in a search session. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

[img]
Preview
PDF
Published Version

Download (3MB) | Preview

Abstract

We study relevance judgment and user activities in a search session. We focus on ephemeral relevance—a contextual measurement regarding the amount of useful information a searcher acquired from a clicked result at a particular time—and two primary types of search activities—query reformulation and click. The purpose of the study is both explanatory and practical. First, we examine the influence of different factors on ephemeral relevance and user activities in a search session. Second, we leverage short-term search history and implicit feedback in a session to predict ephemeral relevance and future search activities. The main findings include:
1. As a contextual usefulness measurement, ephemeral relevance differs from both topical relevance judgment and context-independent usefulness assessment. We show ephemeral relevance significantly relates to a wide range of factors, including topical relevance, novelty, understandability, reliability, effort spent, and search task. The difference between ephemeral relevance and context-independent usefulness assessment is linked to judgment criteria, novelty, effort spent, and changes in user’s perceptions of a search result.
2. Ephemeral relevance can be predicted accurately using implicit feedback signals without any manual explicit judgments. We generalize existing implicit feedback methods from using information related to a single result to those based on user activities in a whole session, achieving a correlation as high as 0.5 between the predicted and real judgments.
3. We show choices of word changes in query reformulation and click decisions significantly relate to recent search history, such as the contents and effectiveness of previous search queries, the contents of the results viewed and clicked in previous searches, etc.
4. Leveraging short-term search history in a session and other information, we can predict word changes in query reformulation and click decisions with different levels of accuracies.
These findings help disclose and explain the dynamics of relevance and user activities in a search session. The developed techniques provide effective support for developing interactive IR systems.


Share

Citation/Export:
Social Networking:
Share |

Details

Item Type: University of Pittsburgh ETD
Status: Unpublished
Creators/Authors:
CreatorsEmailPitt UsernameORCID
Jiang, Jiepujij29@pitt.edujij29
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairHe, Daqingdah44@pitt.eduDAH44
Committee MemberBrusilovsky, Peterpeterb@pitt.eduPETERB0000-0002-1902-1464
Committee MemberLin, Yu-Ruyurulin@pitt.eduYURULIN
Committee MemberKelly, Diane
Date: 24 January 2017
Date Type: Publication
Defense Date: 11 May 2016
Approval Date: 24 January 2017
Submission Date: 19 December 2016
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 264
Institution: University of Pittsburgh
Schools and Programs: School of Information Sciences > Library and Information Science
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: web search, relevance, user interaction, search session, search task, search context
Date Deposited: 24 Jan 2017 20:06
Last Modified: 25 Jan 2017 06:15
URI: http://d-scholarship.pitt.edu/id/eprint/30612

Metrics

Monthly Views for the past 3 years

Plum Analytics


Actions (login required)

View Item View Item