Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.786586
Title: Understanding and predicting user perception of engagement through user behaviour in information retrieval
Author: Zhuang, Mengdie
ISNI:       0000 0004 7972 0333
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
User engagement is an assessment of the quality of user experience while interacting with information systems. Typically, two types of measures are used to assess engagement, one based on user perception, and one on user behaviour. They are either obtrusive to user interaction, or respectively, implicit. Moreover, they have been treated as discrete types of measures and the current body of literature does not suffice to verify their connections. The purpose of this research is to analyse the relationship between user behaviour and user perception in the assessment of engagement in two information retrieval contexts, searching and browsing. In Phase 1, we investigate the role of discrete behavioural features in predicting user perception of engagement through correlation analysis. The resulting predictive model then serves as a baseline and basic framework for phases presented subsequently. In Phase 2, we investigate the added benefits of behaviour sequences through the chi-square test. In Phase 3, based on the findings from our previous phases, we developed and evaluated contextbased measures of engagement. Our measures perform as well as the state-of-the-art approach, without the need for finely-grained data and are interpretable and transferable between different contexts with ease. Findings confirm that a relationship exists between user behaviour and user perception of engagement. Three behaviour patterns that are indicative of engaged/disengaged users were identified. Along the way, we contrast the two information retrieval contexts, drawing parallels and shedding light on particular behaviour patterns only apparent in one or the other. This thesis contributes to a greater understanding of measuring engagement in information retrieval. For the first time, we demonstrate how user behaviour is correlated with user perceived engagement in the context of searching and browsing. Furthermore, we extract easily computable and interpretable measures of engagement, which contribute to the bottom-up methodological approach for measure design.
Supervisor: Toms, Elaine ; Simpson, Andrew Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.786586  DOI: Not available
Share: