From conceptual "Mash-ups" to "Bad-ass" blends: A robust computational model of conceptual blending

Abstract

Conceptual blending is a cognitive phenomenon whose instances range from the humdrum to the pyrotechnical. Most remarkable of all is the ease with which humans regularly understand and produce complex blends. While this facility will doubtless elude our best efforts at computational modeling for some time to come, there are practical forms of conceptual blending that are amenable to computational exploitation right now. In this paper we introduce the notion of a conceptual mash-up, a robust form of blending that allows a computer to creatively re-use and extend its existing common-sense knowledge of a topic. We show also how a repository of such knowledge can be harvested automatically from the web, by targetting the casual questions that we pose to ourselves and to others every day. By acquiring its world knowledge from the questions of others, a computer can eventually learn to pose introspective (and creative) questions of its own.