Ευχρηστία διαδικτύου: σχεδιασμός ιστοτόπων με βάση γνωσιακά μοντέλα διαδραστικής αναζήτησης πληροφορίας

This thesis addresses the problem of effective design of websites that enhance findability of information during interactive search. The term interactive search is used to describe the exploratory, goal−directed browsing activity of a user, while seeking information in a website. Seeking information using a search engine is not an object of study in this thesis. The thesis falls into the field of human computer interaction, and the subfields of Web usability and human information interaction. A website should be treated both as a user interface as well as a hypertext information space, whose structural design has a significant influence on intuitive access to content, task completion and overall user experience. In today’s information explosion, one of the biggest challenges in web design is the effective design of its information architecture. The latter constitutes the research objective of this thesis. The thesis proposes a new methodology to design and evaluate the information architecture of websites, the MEDIAMIS (Methodology for the Efficient Design of Information Architectures based on Models of Interactive Search). The contribution of the proposed approach is that is optimizes and systemizes the information architecture design or evaluation process. MEDIAMIS is inspired by established user-centered techniques and recent cognitive models of user’s interactive search behavior. MEDIAMIS is addressed to Web practitioners and deals with the design or evaluation of a website’s information structure and labeling system. In the context of the proposed methodology, two innovative tools have been developed. The first tool, AutoCardSorter (Automated Card Sorting Tool), supports the structural design of a website. The second tool, ISEtool (InfoScent Evaluator Tool), facilitates the production of semantically appropriate hyperlink labels for the typical goals of a website. The applicability of the proposed methodology and the effectiveness and efficiency of the associated tools were examined through six experimental studies. These studies compared both quantitavely and qualitatively the results of the proposed methodology against the ones derived from established user-centered design approaches, such as card sorting, and against observed and subjective user data. It was found that the proposed methodology lead to a substantial efficiency gain, without expense in the quality of results.