Mark is currently Principal Investigator on the DARPA Integrated Learning Program's POIROT project, where he is coordinating the efforts of fourteen university and industrial research teams to develop a system that can learn hierarchical task procedures or 'workflows' from observations of semantic web service traces.

Dr. Mark Burstein was previously the Director of the Human-Centered Systems Group in BBN's Intelligent Distributed Computing Department, a group of roughly 35 people that includes scientists in the areas of Artificial Intelligence, Psychology, Human Factors, Mixed-Initiative Agent-based Systems, Intelligent Training Systems, and interactive planning and scheduling systems.

Mark's research interests include automated and mixed-initiative planning and scheduling, mixed-initiative agent-based organizations, organizational simulation systems, machine learning, models of human memory organization and retrieval, and cognitive models of plausible analogical and qualitative reasoning. He has a long list of published papers, articles and book chapters.

Mark was a founding member of the OWL-S Coalition, a group of researchers from across the country that worked together to develop the OWL-S, a semantic web ontology and methodology for dynamic utilization of web services. He designed many key aspects of the OWL-S model. (See publications on Semantic Web Services.)

Mark also co-chaired and was the primary organizer of the Semantic Web Service Initiative's Architecture Committee (SWSA), an international group of researchers within the Semantic Web Services Initiative (SWSI) that was chartered to develop an architectural model for web services that can dynamically interoperate with software clients based on published semantic representations of their functionality. He was primary author and editor of the committee report.

From 2001-2005, Mark worked with Prof. Drew McDermott of Yale and Dr. Doug Smith at Kestrel Institute on a project to develop methods for automatically translating OWL-based semantic representations of data between different ontologies that were designed to capture largely the same content. (See publications on Ontology Translation.) It was as a result of this work that he became heavily involved with the OWL-S effort to develop an ontology for modeling semantic web services.

From 1998-2003, Dr. Burstein was the Principal Investigator for a research contract in DARPA's Control of Agent-Based Systems Program (CoABS), under which BBN developed mixed-initiative techniques for the construction and management of software agent teams integrated within human organizations. A major focus of this effort was a large-scale demonstration of mixed-initiative command and control of simulated military organizations. Dr. Burstein led the working group on Mixed-initiative Agent Team Administration (MIATA) comprised of eight research groups within the program to develop a large-scale demonstration of mixed-initiative control agent technology. This group has developed a demonstration using simulation to model the agents and organizations that took part in the relief effort following Hurricane Mitch in Honduras.
(See publications on Mixed-Initiative Agents.)

Under both CoABS and the the DARPA DAML program, Mark collaborated with Prof. Drew McDermott of Yale University and Dr. Doug Smith at Kestrel Institute to develop formal techniques for generating programs to do semantic translations between different onotologies and data formats to enable software agents developed for different purposes to communicate. This is a key challenge for information integration systems. Under the DAML program, Dr. Burstein is also member of a coalition developing an ontology for describing web services, along with researchers at SRI, Stanford and CMU.
(See publications on Agent Ontologies.)

Dr. Burstein also has done work applying and transitioning advanced constraint-based scheduling techniques to Air Force scheduling problems, in collaboration with the Dr. Stephen Smith of the CMU Robotics Institute and Dr. Doug Smith at Kestrel Institute. Here, he developed several scheduling tools as part of the Consolidated Airlift Mobility Planning System (CAMPS) at the Air Mobility Command. This work began in 1994 as joint work with Dr. Doug Smith at Kestrel Institute that resulted in the development of ITAS, an In-Theater Airlift Scheduler.
(See publications on Planning and Scheduling.)

From 1990-1995, Dr. Burstein was technical leader for a DARPA and Air Force Research Laboratory funded project in the DARPA/Rome Planning Initiative (ARPI). Here, he led the development of the Common Prototyping Environment (CPE), a distributed agent-based platform for AI researchers and contractors developing tools for automated planning and scheduling. This environment supported the distributed, collaborative development of new planning and scheduling tools using shared software and data, and the transition of research and development products into working prototypes of tools to assist joint military planners. It was the first major application of the KQML agent communications language. The CPE, under Dr. Burstein's coordination, demonstrated the interoperation of seven different planning and scheduling systems, and included contributions from twelve different organizations. He also proposed and was involved in a number of Technology Integration Experiments with other researchers on the program.
(See publications on Planning and Scheduling)

In 1989, Dr. Burstein worked on several projects including a mixed-initiative interface for a Naval planning tool and a semantic indexing system for documents.

From 1985-1988, Dr. Burstein was technical lead on the BBN Laboratories Knowledge Acquisition Project research team, a DARPA Strategic Computing project to develop knowledge acquisition tools for expert system and natural language processing system knowledge bases. That project developed the KREME knowledge-base development and editing system, some of whose principles were adapted by Dr. Doug Lenat for the CyC system.
(See publications on Knowledge Acquisition and Knowledge Representation)

Prior to coming to BBN, Dr. Burstein was a researcher at Yale University's AI Laboratory, where he did pioneering work on models of human analogical reasoning and learning. He has continued this work at BBN, and expanded it, in collaboration with Dr. Allan Collins, one of the founders of the Cognitive Science Society, to cover issues in plausible and case-based reasoning as well.
(See publications on Plausible and Analogical Reasoning)

In his first year at BBNT, Dr. Burstein also developed a prototype of a mixed-initiative expert advisory system for a statistics software package that was sold commercially by the BBN Software Products Corporation.