Opening the World with Active Words and Concept Triggers

In this paper we present a proposal to help bypass the bottleneck of knowledge-based NLP systems having to work under a closed world assumption. We propose ways of reinterpreting static sources as active ones, by allowing a system to create new lexical entries on the fly, and investigate how to create new concepts on the fly. We argue that a computational lexical semantic approach is a sine qua non to work under an open world assumption. More specifically, we show how to create new lexicon entries using lexico-semantic rules and investigate how to create new concepts for unknown words, building a new model to trigger concepts in context.