Knowledge and reasoning for medical question-answering
Pierre Zweigenbaum · 2009
Restricted domains such as medicine set a context where question-answering is more likely expected to be associated with knowledge and reasoning (Mollá and Vicedo, 2007; Ferret and Zweigenbaum, 2007). On the one hand, knowledge and reasoning may be more necessary than in open-domain question-answering because of more specific or more difficult questions. On the other hand, it may also be more manageable, since by definition restricted-domain QA should not have to face the same breadth of questions as open-domain QA. It is therefore interesting to study the role of knowledge and reasoning in restricted-domain question-answering systems. We shall do so in the case of the (bio-)medical domain, which has a long tradition of investigating knowledge representation and reasoning and, more generally, artificial intelligence methods (Shortliffe et al., 1975), and which has seen a growing interest in question-answering systems (Zweigenbaum, 2003; Yu et al., 2005; Demner-Fushman and Lin, 2007; Zweigenbaum et al., 2007).