AbstractBoth the quantity and complexity of course-related information available to students are rapidly increasing on the Web. This potential information overload challenges standard information retrieval models as users find it increasingly difficult to locate relevant information. The education domain is one of the main domains that has been influenced by this problem. Choosing a higher education course at university can be incredibly tedious and extremely complicated for students. A personalised recommendation system can be an effective way of suggesting relevant courses to prospective students. The existing methods which are mainly based on keywords fail to address the individual user’s needs in the recommendation process. Although models use collaborative filtering there is often a lack of historical information. Another shortcoming is that they do not provide comprehensive knowledge about the course that is most relevant to the student.
This research presents a novel ontology-based hybrid approach to recommend personalised courses to match student’s individual needs by integrating all available information about courses and supporting students to choose courses towards their career goals. This thesis makes three major contributions: firstly, it proposes a comprehensive Ontology based Personalised Course Recommendation framework, called OPCR, by combining several artificial intelligence techniques including collaborative filtering, content-based filtering, ontological representation and management of knowledge. A set of ontology based recommendation algorithms are developed for personalised recommendation. The framework is capable of automatic data extraction, integration to provide students with suitable recommendations to meet their needs. It not only reduces information overloading but also improves recommendation accuracy. Secondly, it proposes ontology models to extract and integrate information from multiple sources, which contributes to improving the quality of the recommendations by overcoming the heterogeneity of course information. In addition, it has properties such as generality which enables it to be used in different recommendation system domains which change with the user’s interests and the object’s attributes. Finally, a personalised recommendation system based on the OPCR framework is developed and evaluated. The system is available online as open access for researchers and developers. Results show that the ontology based recommendation algorithms that use hierarchically related concepts produce better outcomes compared to a filtering method that considers only keyword similarity. In addition, the system’s performance is improved when the ontology similarity between the items' profiles and the users' profiles is utilised.
|Date of Award||Jan 2019|
|Supervisor||Linda Yang (Supervisor)|