Motivation Modelling and Computation for Personalised Learning of People with Dyslexia




Journal Title

Journal ISSN



Volume Title


De Montfort University


Thesis or dissertation

Peer reviewed


The increasing development of e-learning systems in recent decades has benefited ubiquitous computing and education by providing freedom of choice to satisfy various needs and preferences about learning places and paces. Automatic recognition of learners’ states is necessary for personalised services or intervention to be provided in e-learning environments. In current literature, assessment of learners’ motivation for personalised learning based on the motivational states is lacking. An effective learning environment needs to address learners’ motivational needs, particularly, for those with dyslexia. Dyslexia or other learning difficulties can cause young people not to engage fully with the education system or to drop out due to complex reasons: in addition to the learning difficulties related to reading, writing or spelling, psychological difficulties are more likely to be ignored such as lower academic self-worth and lack of learning motivation caused by the unavoidable learning difficulties. Associated with both cognitive processes and emotional states, motivation is a multi-facet concept that consequences in the continued intention to use an e-learning system and thus a better chance of learning effectiveness and success. It consists of factors from intrinsic motivation driven by learners’ inner feeling of interest or challenges and those from extrinsic motivation associated with external reward or compliments. These factors represent learners’ various motivational needs; thus, understanding this requires a multidisciplinary approach.

Combining different perspectives of knowledge on psychological theories and technology acceptance models with the empirical findings from a qualitative study with dyslexic students conducted in the present research project, motivation modelling for people with dyslexia using a hybrid approach is the main focus of this thesis. Specifically, in addition to the contribution to the qualitative conceptual motivation model and ontology-based computational model that formally expresses the motivational factors affecting users’ continued intention to use e-learning systems, this thesis also conceives a quantitative approach to motivation modelling. A multi-item motivation questionnaire is designed and employed in a quantitative study with dyslexic students, and structural equation modelling techniques are used to quantify the influences of the motivational factors on continued use intention and their interrelationships in the model.

In addition to the traditional approach to motivation computation that relies on learners’ self-reported data, this thesis also employs dynamic sensor data and develops classification models using logistic regression for real-time assessment of motivational states. The rule-based reasoning mechanism for personalising motivational strategies and a framework of motivationally personalised e-learning systems are introduced to apply the research findings to e-learning systems in real-world scenarios. The motivation model, sensor-based computation and rule-based personalisation have been applied to a practical scenario with an essential part incorporated in the prototype of a gaze-based learning application that can output personalised motivational strategies during the learning process according to the real-time assessment of learners’ motivational states based on both the eye-tracking data in addition to users’ self-reported data. Evaluation results have indicated the advantage of the application implemented compared to the traditional one without incorporating the present research findings for monitoring learners’ motivation states with gaze data and generating personalised feedback.

In summary, the present research project has: 1) developed a conceptual motivation model for students with dyslexia defining the motivational factors that influence their continued intention to use e-learning systems based on both a qualitative empirical study and prior research and theories; 2) developed an ontology-based motivation model in which user profiles, factors in the motivation model and personalisation options are structured as a hierarchy of classes; 3) designed a multi-item questionnaire, conducted a quantitative empirical study, used structural equation modelling to further explore and confirm the quantified impacts of motivational factors on continued use intention and the quantified relationships between the factors; 4) conducted an experiment to exploit sensors for motivation computation, and developed classification models for real-time assessment of the motivational states pertaining to each factor in the motivation model based on empirical sensor data including eye gaze data and EEG data; 5) proposed a sensor-based motivation assessment system architecture with emphasis on the use of ontologies for a computational representation of the sensor features used for motivation assessment in addition to the representation of the motivation model, and described the semantic rule-based personalisation of motivational strategies; 6) proposed a framework of motivationally personalised e-learning systems based on the present research, with the prototype of a gaze-based learning application designed, implemented and evaluated to guide future work.





Research Institute