Monday, 8:30

 

Exploring the Relationships Between Search Intentions and Query Reformulations

Nicholas J Belkin1, Eun Youp Rha1, Matthew Mitsui2, Chirag Shah1
1Rutgers University, School of Communication & Information, United States of America; 2Rutgers University, Department of Computer Science

Summary

We report on a study investigating the relationships among query reformulations and different search intentions during an information seeking session. 24 participants were each asked to search for information useful for two (of four) different journalism tasks; after completing each search, the search was replayed, and participants were asked to specify what they intended to accomplish in each query segment of the search session, and whether those intentions were satisfied. Logs of the searches were analyzed to extract the queries at the start and finish of each query segment, and query reformulations were classified. Results show that: participants regularly indicated a variety of different search intentions during the course of an information seeking session; there are some differences in reformulation types following different search intentions; there are some differences in reformulation types which follow satisfied and unsatisfied intentions; and, there are differences in the frequency of intentions following reformulations which themselves follow satisfied and unsatisfied intentions. Implications for system design are discussed.


The Exploration of Objective Task Difficulty and Domain Knowledge Effects on Users’ Query Formulation

Chang Liu1, Xiangmin Zhang2, Wei Huang1
1Peking University, China, People’s Republic of; 2Wayne State University

Summary

In this paper, we explore the effects of objective task difficulty and domain knowledge on users’ query formulations. The dataset from a user experiment was used in this research, which focused on the medical domain. The objective difficulty was measured by the precision level of search topics (as queries) in the search system, and searchers’ domain knowledge was assessed by a self-reported rating on the familiarity with MeSH terms that were related to the search topics in the study. We compared expert searchers’ and novice searchers’ query similarity and query features between easy and difficult tasks. The results showed that there was no significant difference between domain experts and novices in query similarity, but there existed an opposite pattern of task difficulty on query similarity for searchers with different domain knowledge levels. Domain expert searchers had more diverse vocabulary in difficult tasks than in easy tasks, whereas novice searchers had to rely heavily on task descriptions to formulate their queries. Task difficulty had also influenced searchers’ performance, especially the precision measure. Novice searchers’ recall was relatively low in both easy and difficult tasks. The findings in this study help us further understand users’ query formulation process, and have implication for the design of query suggestion functions in search systems.


What Makes a Query Temporally Sensitive?

Craig Willis, Garrick Sherman, Miles Efron
University of Illinois at Urbana-Champaign, United States of America

Summary

This work examines factors that affect manual classifications of “temporally sensitive” information needs. We introduce the concepts of “temporal relevance” and “temporal topicality” to differentiate between different aspects of temporal retrieval research. We use qualitative and quantitative techniques to analyze 660 topics from the Text Retrieval Conference (TREC) previously used in the experimental evaluation of temporal retrieval models. We use regression analysis to model previous manual classifications. We identify factors and potential problems with previous classifications, proposing principles and guidelines for future work on the evaluation of temporal retrieval models.