The use of contextual information in a proactivity model for conversational agents

For conversational agents engaging in a natural language-based interaction with web site users in service exchange applications, simple entertaining "chatting" is not sufficient. Instead, the agent needs to be cooperative by trying to provide relevant information about products and/or the conditions of purchasing. In this paper we analyze how the proactive behaviour of conversational agents can be used to increase the general user satisfaction. The results of two case studies allow us to outline a generic proactivity, model for information agents based on retrieval mechanisms and search heuristics.