Conversational Agent that Models a Historical Personality

In this paper we discuss the current approaches in question answering (QA) and their applicability in building a conversational agent that models a historical figure that gives informative and relevant answers to user questions about the life of that personality. We analyze two main methods: one in which we use an ontology to build our knowledge base and one where we don't have a knowledge base and we solve the answer sentence selection problem for question answering. We observed that the first method is better for answering more general questions and the second method can deal with more specific and complex lexically and syntactically questions. The implementation of the conversational agent relies on the two combined approaches, the second being a fallback option if the first method is not able to provide an answer. Author