A knowledge-based framework for the rapid development of conversational recommenders

Web-based sales assistance systems are a valuable means to guide online customers in the decision-making and product selection process. Conversational recommenders simulate the behavior of an experienced sales expert, which is a knowledge-intensive task and requires personalized user interaction according to the customers' needs and skills. In this paper, we present the ADVISOR SurrE framework for rapid development of conversational recommenders for arbitrary domains. In the system, both the recommendation logic and the knowledge required for constructing the personalized dialog and adaptive web pages is contained in a declarative knowledge-base. The advisory application can be completely modeled using graphical tools based on a conceptual model of online sales dialogs. A template mechanism supports the automatic construction of maintainable dynamic web pages. At run-time, a controller component generically steers the interaction flow. Practical experiences from several commercial installations of the system show that development times and costs for online sales advisory systems can be significantly reduced when following the described knowledge-based approach.