WatsonPaths: Scenario-based Question Answering and Inference over Unstructured Information

We present WatsonPaths^TM , a novel system that can answer scenario-based questions, for example medical questions that present a patient summary and ask for the most likely diagnosis or most appropriate treatment. WatsonPaths builds on the IBM Watson^TM question answering system that takes natural language questions as input and produces precise answers along with accurate confidences as output. Watson-Paths breaks down the input scenario into individual pieces of information, asks relevant subquestions of Watson to conclude new information, and represents these results in a graphical model. Probabilistic inference is performed over the graph to conclude the answer. On a set of medical test preparation questions, Watson-Paths shows a significant improvement in accuracy over the base Watson QA system. We also describe how WatsonPaths can be used in a collaborative application to help users reason about complex scenarios.