A Framework for Usage Mining based Usability Analysis

Abstract

The world wide web (www) is a global phenomenon and the in uence of www on business application domains is fundamental in designing and delivering quality on-line products. The quality of an on-line product depends primarily on the correctness of the application, i.e. the application does the job it is designed for. However, the e ectiveness of the application depends on the users' experience with the application, i.e. whether the application allows the user to interact with the application seamlessly. Usability e ectiveness of an application is thus an important step in establishing the success of a web application. Traditionally, the usability analysis of a web application is performed by soliciting responses from the user in the form of lled questionnaires. Then the user responses are analysed to assess the usability of the application. However, whilst sharing the user experience, the user may or may not be able to report the `exact' application behaviour. Web usage mining has been in practice to predict about the future business of organisations and many web usage mining techniques have been proposed in literature. However, web usage mining techniques generally focus on revenue generation, web tra c routing, advertisement placement, etc. by way of analysing the web logs. In this work, we focus on the `automated' usability analysis of a web application using user-interaction patterns. We conjecture that the user-interactions depict the user-experience of the application, i.e. if the users are able to perform the tasks they desire, seamlessly or otherwise, this should be re ected in the userinteraction patterns. We propose a novel approach that works by processing the sequences of interaction obtained from user-interactions with the application and produces a usability analysis of the application similar to the one obtained by lling questionnaires. We use Jakob Nielsens' usability heuristics for the purpose and show that user-interaction mining can be used for usability assessment of a web application. We have used a customized Learning Management System for this study that records user-interactions and then use state-of-the-art sequence mining technique based on pattern-growth paradigm to e ciently extract the sequences that could help in establishing the usability e ectiveness of the application based on Nielsens' usability heuristics. The experimental results give a comparison of the results obtained from the traditional questionnaire based approach and the interaction mining based approach we propose and thus, show the usefulness of the interaction mining based usability analysis.