Using terminological feedback for web search refinement

Abstract

Although interactive query reformulation has been actively studied in the laboratory, little is known about the actual behavior of web searchers who are offered terminological feedback along with their search results. We analyze log sessions for two groups of users interacting with variants of the AltaVista search enginea baseline group given no terminological feedback and a feedback group to whom twelve refinement terms are offered along with the search results. We examine uptake, refinement effectiveness, conditions of use, and refinement type preferences. Although our measure of overall session "success" shows no difference between outcomes for the two groups, we find evidence that a subset of those users presented with terminological feedback do make effective use of it on a continuing basis.

Differences between web sessions according to the origin of their visits

“…The first studies were focused in the functionality of the search engines, describing query patterns in Altavista (Silverstein, Henzinger, Marais, & Moricz, 1998;Silverstein, Marais, Henzinger, Moricz, 1999;Anick, 2003), Excite and Alltheweb (Jansen & Spink, 2006) and Yahoo! (Teevan et al, 2006).…”

Web log analysis: a review of a decade of studies about information acquisition, inspection and interpretation of user interaction

“…Several Web search engines have begun offering short lists of search refinement suggestions to encourage the interactive narrowing of query result sets. Anick (2003) reports on the results of several log-based studies of user interaction with the AltaVista 4 Prisma assisted search tool. In particular, they study the degree and nature of user uptake of the feature, as well as the conditions and effectiveness of its use within information seeking sessions based on an analysis of anonymous user activity logs from the AltaVista search site.…”

“…Among all the papers reviewed in this work, only a few explicitly claim the need of a combination of different sources of evidence: Mahoui and Cunningham (2000) write that further studies, including interview-based examinations of small groups of users, are required to complete their work; Anick (2003) needs laboratory tests or questionnaires to answer more complex questions. Koch et al (2004) found that a thorough log analysis can provide deeper understanding of the services and they may offer useful hypotheses for advanced user studies.…”

Implementing and evaluating phrasal query suggestions for proximity search

“…In spirit, our super phrases are similar to those generated by Anick from top-ranked documents and past queries in the Prisma system [2], and by Gauch from a back-of-the-book index in an expert system for searching [18]. The hyperindex meta-search engine described by Bruza and Dennis [7] appears to be closest to our approach of narrowing and widening.…”

“…2 For each of these subphrases there are two orderings. The subphrase AB is covered by four minimal sequences in d:…”