Statistical Approaches to Question Answering in Watson

Introduction The ability to understand and communicate in natural languages, such as English or Chinese, is a hallmark of human intelligence, and one of the core challenges in the field of Artificial Intelligence. Despite decades of AI research, the goal of building machines with human-level conversational fluency, as depicted by HAL and C3PO in science fiction, remains elusive and daunting. Nevertheless, a burst of recent progress in the field of Natural Language Processing (NLP) has been enabled by the explosive growth in machine-readable language data (primarily in text or hypertext form) available from sources such as the World Wide Web. Furthermore, increasingly powerful computer hardware makes it feasible to apply complex analysis to these large volumes of text. Consequently, researchers have made exciting progress in automated Question Answering (QA), which aims to find a specific answer to a user’s natural language question.