VISUAL WEB MINING FOR WEBSITE EVALUATION

In this paper we present an interactive system named Website Exploration Tool (WET) that aims at supporting the evaluation of websites through the exploration of web data. Our prototype offers a set of coordinated visual abstractions in the form of interactive graphs and trees to analyse data graphs within meaningful contexts such as the website structure and users' flow. Apart from classical approaches, our highly interactive system introduces a wide variety of information visualisation techniques that provide visual cues to assist in the process of digging into usage data. Among them, we present a new hierarchical approach for characterising users' flow which simplifies the intricate graphs generated by real users browsing, and a technique for extracting contextual subgraphs simplifying the task of visualising very large websites. The interface of the system has been evaluated with expert analysts that validated the usefulness of the tool for analysing a wide variety of websites.