Usability and User Experience Information in Reviews – Københavns Universitet

Usability and User Experience Information in Reviews

PhD defence by Steffen Hedegaard

Abstract

Product reviews available on-line contain information on users’ interaction with the product in the form of narratives, describing the users’ experiences with using the product in the wild. The large amount of reviews available makes it a potential interesting source of information related to product use.

This thesis works toward answering the following research questions about the information available in reviews:

1. How do users describe their interaction with products in narratives such as product reviews?

2. Can such information be used to improve future versions of the product?

Results show that users do write about product use in terms related to standard and popular aspects of usability and user experience, e.g. efficiency, effectiveness, enjoyment, frustration. The frequency with which different aspects are depicted in reviews differ significantly between product domains. We also find that reviews can be a source of information with regard to identifying more persistent usability issues. In both of the two cases mentioned above, we device automatic methods for classifying sentences with regard to such content.

A comparison of supervised machine learning algorithms with semi supervised learning (SSL) algorithms on several standard corpora for text classification show that support vector machines normally are the best choice for text classification. We also find that traditional feature vectors consisting of counts of function words can be used for identifying the author of a translated document but the method can be augmented by adding semantic information from documents.

Overall the results presented in this thesis help clarify the role of reviews in relation to understanding both the users and different aspects of product use.

Assessment committee:

Chairman: Professor Kasper Anders Søren Hornbæk, Department of Computer Science, University of Copenhagen, Denmark,

Professor Kaisa Väänänen-Vainio-Mattila, Professor, Human-Centered Technology (IHTE), Dept. of Pervasive Computing,

Tampere University of Technology, Finland

Professor Michael Terry, David R. Cheriton School of Computer Science, University of Waterloo, Canada

Academic Supervisor:

Professor Jakob Grue Simonsen, Department of Computer Science, University of Copenhagen, Denmark

For an electronic copy of the thesis, please contact Marianne Henriksen, marianne@di.ku.dk.