Thesis Title
Where do queries come from? An eyetracker study of how users are influenced to search

Research Description
While research has proven that query variations exist, not much work has investigated why these variations exist. Our aim is to gain a better understanding of query formulation process and to get an insight into where query variations are coming from.

Building on previous research we will be investigating whether there is relationship between eye fixation on specific words users came across prior to starting their search session and the queries they formulated. We will be using the UQV100 test collection because it takes the participant’s variation and query variation into consideration. Where participants will be wearing eye trackers while looking at the backstories, then we will map down the eye fixation to the individual word’s level. Participants will be asked to type their queries and rank a set of query variations based on the query order they perceive to generate the most useful results.

We will also investigate whether fixation on words from the backstories and participants curiosity levels have any relationship with the queries they type, how well they can type those queries and rank the set of the given query variations. For measuring curiosity level, we will be using self-reported curiosity ratings and machine learning algorithm to analyse eye movement pattern.

Then we will be evaluating the typed queries and the participant’s query ranking results with the actual usefulness of the query via Ranked-Biased-Overlap (RBO) and Pearson’s correlation against NDCG@10 scores.

By understanding how users formulate their queries, we can get an insight into where query variations come from, which can assist the research community in designing search engines or information retrieval systems that can accommodate users’ diversity, as it is important for users with varying levels of expertise to be able to utilise search systems successfully to retrieve information ensuring equity when searching and retrieving information.

Supervisors
Prof Mark Sanderson, RMIT University
Prof Dr Dana McKay, RMIT University