The interplay of language processing, reasoning and decision-making in cognitive computing

Abstract

Integrating language processing, reasoning and decision making is a prerequisite to any breakthroughs in cognitive computing. This paper discusses historical attitudes that have worked against such integration, then describes a cognitive architecture called OntoAgent that illustrates both the feasibility and the payoffs of pursuing integration. Examples are drawn from the Maryland Virtual Patient prototype application, which offers medical trainees the opportunity to diagnose and treat a cohort of cognitively modeled virtual patients that are capable of language processing, reasoning, learning, decision making and simulated action.