Diagnosing Faulty Transitions in Recommender User Interface Descriptions

Abstract

Complex assortments of products and services offered by online selling platforms require the provision of sales support systems assisting customers in the product selection process. Knowledge-based recommenders are intelligent sales assistance systems which guide online customers through personalized sales dialogs and automatically determine products which conform to their needs and wishes. Such systems have been successfully applied in a number of application domains such as financial services or digital cameras. In this context, the construction of recommender user interfaces is still a challenging task. In many cases faulty models of recommender user interfaces are defined by knowledge engineers and no automated support for debugging such models is available. In this paper we discuss a formal model for defining the intended behaviour of recommender user interfaces and show the application of model-based diagnosis concepts which allow the automated debugging of those definitions. Experiences show that this approach significantly increases the productivity of recommender user interface development and maintenance.