Knowledge-based support for evaluation of computer-based products

Modern computer-based products are actively finding new applications in our daily lives. Smartphones, vehicles and home appliances are getting "smarter" in their effort to provide more assisting and entertainment services to their owners. From the development perspective this often requires significant financial investments and inevitably brings up a number of efficiency-related questions. For instance: is the new functionality well-accepted by the users or is it just overloading the product? Should all functions be further supported or have some of them become obsolete? Are there some goals that the users cannot achieve with the product? Is the overall product perception positive or is there a room for improvement? In order to answer such questions and to tailor next product versions to what their users need, an extensive and comprehensive evaluation of product quality has to be performed. In case with complex devices this is a knowledge-intensive process that requires considerable human expertise and effort. Identification of relevant data sources, their processing and correct interpretation is challenging. For example, if a function is called less often than expected, does this indicate that it is non-compliant to the user needs? Which aspects may have caused this situation? How can user opinion or other sources be incorporated to reveal these causes? We claim that information technologies are capable of helping developers to infer the required knowledge from various data on product usage. An introduction of a knowledge-based system can support the evaluation process and open up new possibilities for its efficiency increase. In this thesis we present a knowledge model to establish a basis for such systems. The model covers the user and the developer views on the products being in operational use. This includes generic knowledge on the user and the product, as well as knowledge that can be acquired during their interaction and can be used to improve the product. The constructed model is well-suited for building a knowledge-based system upon it. In order to illustrate this, we create a system prototype using ontological technologies and demonstrate it on an in-vehicle infotainment system. The prototype receives user feedback and data on the product usage and suggests aspects that might be deviating from the user needs. The thesis is wrapped up with a conclusion and an outlook into future research.