This week in the Syn/Sem Discussion Group, we will have a presentation by Marisa Boston, a recent Cornell Linguistics PhD, and now Senior Research Engineer in Natural Language Processing a Nuance Communications.
Syntax and semantics in WatsonMD
The super-computer Watson made headlines when it outperformed the reigning Jeopardy! champions in 2011. The Jeopardy! game-show represents a particularly challenging problem to artificial intelligence, and question-answering in particular, because it is open-domain: questions and answers come from a wide variety of topics, and systems require inference to make connections between often disparate topics. To do this, Watson goes beyond simple bag-of-words techniques to provide deeper syntactic and semantic analysis of questions and answers.
While the Jeopardy!-winning Watson represents an incredible feat in artificial intelligence, it has only begun to address the challenges of making question-answering systems practical problem-solving tools within and across domains. This talk describes how Nuance and IBM are extending the Watson technology to create a medical question-answering tool for doctors. In particular, I address the syntactic and semantic challenges we face as we make the transition from Watson to WatsonMD.