摘要：Controlled Natural Languages (CNLs) are efficient languages for knowledge acquisition and reasoning. They are designed as a subset of natural languages with restricted grammar while being highly expressive. CNLs are designed to be automatically translated into logical representations, which can be fed into rule engines for query and reasoning. In this work, we build a knowledge acquisition machine, called KAM, that extends Attempto Controlled English (ACE) and achieves three goals. First, KAM can identify CNL sentences that correspond to the same logical representation but expressed in various syntactical forms. Second, KAM provides a graphical user interface (GUI) that allows users to disambiguate the knowledge acquired from text and incorporates user feedback to improve knowledge acquisition quality. Third, KAM uses a paraconsistent logical framework to encode CNL sentences in order to achieve reasoning in the presence of inconsistent knowledge.
Loading...
