Decision making in Web searching: what do searchers look at and why?

This project will investigate, through a series of observational and experimental studies, the decisions made by people searching the Web. Specifically we are interested in how the features of Web pages, such as the textual content or the visual presentation of pages, are used to make interactive search decisions and how the construction of Web pages influence a searcher's decisions about what pages are useful to their search. This study will exploit the relatively novel investigative technique of eye-tracking which allows us to analyse search behaviour by detecting how a searcher visually assesses information. Modern network speeds and the response time of Web search engines facilitate very short interaction cycles in which searchers make very quick interactive decisions after viewing a very small sample of the information available. So how does the availability of large amounts of information that can be processed very quickly change how people assess information? What type of decisions do people make when searching the web and how are these decisions influenced by the content of web pages? Knowing more about how search decisions are made means that we can better support the process of searching: designing more useful surrogates, designing retrieval systems that incorporate more information than simply text or designing more sophisticated querying interfaces for searching.

This project sought to better understand how people make decisions about information presented on the Web, concentrating specifically on how features of Web pages, such as textual content and layout, influence a searcher's decision about what pages are useful to their search. Our user study involved the completion of questionnaires, use of eye tracking technology, talk aloud protocols and post-search interviews. As opposed to previous studies, the study asked participants to search for real and simulated information needs that represented different search contexts (e.g. from searches about personal interest to academic related searches). This permitted the identification of several relevance criteria that naturally occur across different search contexts and the emergence of some fixation patterns, not observed before, associated to the use of these criteria. Through a detailed examination of searcher behaviour we were able to particularly able to shed light on how people use search engine interfaces to make predictions about what information might be useful for their information needs.