The Rachelbot: Investigation of Persona-Based Chatbots. In Data Science & Engineering Master of Advanced Study (DSE MAS) Capstone Projects

Creating agents that closely mimic human conversational patterns has been a major goal of research and industry for decades. Recent advances in recurrent and transformer deep learning architectures have made significant progress toward open domain chatbots that capture the highly contextual nature of conversations. In the last several years, researchers have turned to personas as one way of improving dialogue quality.

In this project, current methods for building persona-based conversational agents are tested using a chatbot of a popular character (Rachel from the TV show Friends). Our aim is to: 1) deliver an engaging Friends experience to fans of the show, 2) assess chatbot deep learning architectures, and 3) establish the limits of contemporary approaches and potential areas for additional research.