17 June 2022. 13:00h. Online session. Language: English.
Exploration-Exploitation trade-off in websearch: effects of search context and prior domain knowledge
Cheyenne Dosso
University of Toulouse Jean-Jaurès
The information searching is a complex cognitive activity where users try gaining much information in a minimum of time, without spending too many resources. From this perspective, users implement different search strategies: Exploration (i.e. processing a larger part of the search space) versus Exploitation (i.e. processing a specific part of the search space).
Furthermore, current search engines such as Google are effective to support users when they are performing simple tasks (e.g. fact-finding), but they do not sufficiently support users in more complex tasks leading to learning or requiring the processing of multiple search paths. A better understanding of how users solve the exploration-exploitation trade-off when performing a complex task would allow improving search engines. For example, search engines could better support the exploration-exploitation trade-off by helping users gain much information while reducing the cognitive cost of information search.
In this talk, I will introduce the cognitive model of information search and I will review the different ways to operationalize the exploration-exploitation strategies. Then, I will present two main factors affecting them: the search context and the level of prior domain knowledge. Finally, I will present some preliminary results from a study using Eye Tracking to identify the exploration-exploitation trade-off implemented by users with or without domain knowledge in different search contexts (i.e. Lookup versus Exploratory).
Bio
I am a PhD student in Cognitive psychology applied to ergonomics at the University of Toulouse Jean-Jaurès (CLLE lab). My thesis is part of the multidisciplinary ANR-CoST project aiming at describing users' behavioral patterns and search strategies when they are solving complex search tasks (https://www.irit.fr/COST/consortium/). My main research interests include Search as Learning, search behavior, expertise, complex activity and tasks.
