| Summary: | The exploitation of the underlying semantics of data inherent in the vision of the Semantic Web tackles the limitations of the traditional keywords-based retrieval model and has the ability to change the way search is done. The proliferation of Open Data published on the Web in recent years has driven significant research and development in search. As a result, there is a wide range of approaches with respect to the style of input, the underlying search mechanisms and the manner in which results are presented. Although the performance or effectiveness of these approaches is usually evaluated, understanding their usability and suitability for end users' needs and preferences has been largely overlooked. This is the main motivation behind the work presented in this thesis. The thesis, thus, presents different pieces of work in this area. The �first part focuses on investigating the usability of different query approaches from the perspective of expert and casual users through a user-based study. The �findings of this study show the strengths of graph-based approaches in supporting users during query formulation with a drawback of high query input time and user effort. Therefore, in another user-based study, learnability of a graph-based approach is evaluated to assess the effects of learning and frequency of use on users' proficiency and satisfaction. The results of both studies suggest that the combination of a graph- based approach with a NL input feature could provide high level of support and satisfaction for users during query formulation. This is, hence, the third piece of work presented in the thesis: a hybrid query approach together with a user-based evaluation to assess its usability and users' satisfaction. The thesis also presents thorough analysis of state-of-the-art in semantic search evaluations and describes a set of best practices for running them based on this analysis, lessons learnt from the Information Retrieval community and on my own experience in evaluating semantic search approaches.
|
|---|
