Lifelong and Continual Learning Dialogue Systems

The purpose of writing this book is to introduce the emerging topic of lifelong/continual learning dialogue systems. Dialogue systems, also commonly known as chatbots, are computer programs that can converse with humans to perform intended tasks. They typically employ text or speech modes. As deep learning has improved the accuracy of speech recognition, text generation, and processing dramatically since 2012 or so, dialogue systems are becoming increasingly ubiquitous and are used in all types of applications such as smartphones, cars, home appliances, company websites, and mobile robots. They provide a very rich set of services, e.g., performing some specific tasks and chit-chatting with human users. However, the user experiences have not been very satisfactory so far (discussed in Chap. 1). It is clearly unfair to expect a deployed dialogue system to be able to understand everything that users may say, but these systems should be able to learn during conversations by interacting with users to become more and more knowledgeable and powerful. That is the goal of building lifelong learning dialogue systems. That is also
the key motivation for us to write this book to introduce and promote the research of lifelong learning dialogue systems.