E-Research Roundtable - Modeling User Searching Behaviors and Search Assistance Usage via Transaction Logs
Description: In order to optimize search and discovery services, it is important to develop evidence-based models of user information seeking behaviors within distributed retrieval environments. While a large number of user information seeking studies have been performed, our knowledge of user searching patterns, particularly in online catalogs (OPACs), is incomplete and often contradictory. The University of Illinois at Urbana-Champaign Library has been collecting custom transaction log data from their main gateway interface and its underlying Easy Search (ES) federated search system since 2007. ES provides contextual and adaptive search assistance mechanisms that present the user with search modification and reformulation suggestions and perform additional target searches in the background. The Illinois team performed a detailed analysis of the project’s custom transaction logs collected over the Fall 2010 and Spring 2011 semesters. This analysis looked at approximately 1.4 million user searches and over 1.5 million user target clickthroughs. This analysis has revealed rich information on user search characteristics, search assistance usage, and user clickthrough actions. This transaction log analysis provides several implications for web-scale discovery system design.
Among the findings: users of the Illinois gateway enter an average of 4.33 terms per search query – much higher than previous studies; 48.05% of the search sessions contain more than one search term or a combination of search terms and search assistance actions – also higher than other studies; and while 66% of all searches originate as default keyword searches, the percentage of known-item or specific title/author searches exceeds 51% of the search queries. Known-item searches are performed in almost 55% of the search sessions. In addition, the ES search assistance suggestions and custom links are well-accepted by users; in 32.45% of all search sessions and 58% of the sessions with more than a single search query, users employed one or more search assistance operations. The logs also revealed that users are entering complete or partial journal titles and then clicking through at a high frequency into an A-to-Z e-journal list link and that the exact phrase/title words added links shown in selected results displays are heavily used. Users click on the presented journal title link 21.41% of the time that they are suggested and in over 6.86% of all search sessions. In addition, the journal title search option tab constitutes over 12% of the searches within the gateway. The use of publisher e-book matches is also high -- with clickthroughs into all the e-book content targets totaling 9.31% of all result target clicks and taking place in 11.36% of all search sessions.
This study will be published shortly. Full citation and an advance copy is posted here, under Resources.
Resources:William H. Mischo, Mary C. Schlembach, Josh Bishoff, Elizabeth German, "User Search Activities within an Academic Library Gateway: Implications for Webscale Discovery Systems", in Planning and Implementing Resource Discovery Tools in Academic Libraries, edited by Mary Popp and Diane Dallis, IGI Global, 2012 (in press), 22 ms pages.