Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.657279
Title: Modelling students' behaviour and affect in ILE through educational data mining
Author: Mavrikis, Manolis P.
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 2008
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Abstract:
The long-term objective behind the research presented in this thesis is the improvement of ILEs and particularly those components that take into account students’ behaviour, as well as emotions and motivation. In related research, this is often attempted based on intuition, theoretical perspectives, or guided by results from studies in the isolation of a research lab. In this thesis, an attempt is made to inform the design of adaptation and feedback components by collecting and analysing as realistic data as possible. Guided by the belief that qualitative data analysis results can be enriched by employing statistical and machine learning techniques, the focus of this research is to investigate (a) key aspects of students’ behaviour and their relation to their learning and (b) how their behaviour could be employed to predict students’ affective and motivational characteristics. The first step is to gain an in-depth understanding of students’ behaviours in ILEs when they interact on their own time and location, rather than during a study where the social dynamics are different. Based on results, components of an ILE are redesigned and two Bayesian models are machine-learned; one that predicts when students need help in answering a question and one that predicts if their interaction with the system is beneficial to their learning. In the next step, machine learning is employed in order to derive predictive models of students’ affective and motivational states based on their interaction with ILEs. This is achieved by deriving decision trees based on a dataset of students’ self-reports collected during replays of their interaction. In addition, in order to take tutors’ perspectives into account, two different approaches are followed: The first attempts to elicit tutors’ inferences while they are watching replays of students’ interactions. This was not entirely successful. In the second approach, decision trees are derived from a dataset of tutors’ inferences collected during one-to-one computer-mediated tutorials.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.657279  DOI: Not available
Share: