Dr. Sergei Nirenburg, professor, specializes in cognitive systems and natural language processing. He is a member of the truly smart agents research group (TSA), a team made up of members of UMBC’s Institute for Language and Information Technology (ILIT), and colleagues from the University of Maryland School of Medicine. TSA was conceived in 2005, and since then has focused on creating artificial intelligent agents that exhibit human-like behavior.
Research within the truly smart agents groups falls within four areas: Ontological Semantics, Cognitive Modeling, Physiological Simulation and Machine Learning. In his paper entitled “Ontological Semantics,” Dr. Nirenburg describes Ontological Semantics as “an integrated complex of theories, methodologies, descriptions, and implementations, attempts to systematize ideas about both semantic description as representation and manipulation of meaning by computer program.” Cognitive Modeling involves building intelligent agents that replicate human behavior, while Physiological Simulation deals with simulating physiological functioning of these agents using scripts. In terms of Machine Learning, the TSA is using bootsrapping methods to expand the OntoSem lexicon and ontology.
Dr. Nirenburg has served as the director of UMBC’s Institute for Language and Information Technology (ILIT) since 2002. ILIT is a research center that combines research in language processing with the development of practical information technology applications and knowledge acquisition tools. Some special tools used at ILIT to pursue intelligent agent research are DEKADE, the Development, Evaluation, Knowledge Acquisition and Demonstration Environment for OntoSem text processing and REDEE, the Reference Engine Development and Evaluation Environment.
Dr. Nirenburg explains that his research goal is two-fold. First, he hopes to develop a symbolic explanatory theory of human cognitive functioning: multiple channels of perception; mixed rational and irrational, affective decision making; short and long-term memory management; and simulated physical and mental and actual verbal action. Secondly, he hopes to build artificial intelligent agents that can fill roles currently played by people in useful applications—a goal that has been realized via his Maryland Virtual Patient (MVP) project.
According to the truly smart agent’s website, MVP is a “multi-level heterogeneous agent-oriented environment” that simulates a doctor-patient relationship. Within this environment, humans, who play the role of doctor, interact with a “virtual patient.” The virtual patient models and simulates both the physiological and the cognitive functionality of a human, while undergoing both normal and pathological processes in response to internal and external stimuli. The virtual patient experiences symptoms, remembers things, and relays its symptoms to the human “doctor.” Dr. Nirenburg has been working on MVP since 2006 and hopes for it to become a helpful learning tool in the medical sphere.