General
Project Halo is a staged research effort by Vulcan Inc. towards the development of a “Digital Aristotle,” a staged, long-term research and development initiative that aims to develop an application capable of answering novel questions and solving advanced problems in a broad range of scientific disciplines. The Digital Aristotle is being developed with a focus on two primary functions: as a tutor capable of instructing and assessing students in the sciences, and as a research assistant with broad, interdisciplinary skills to help scientists in their work.

The Halo Pilot
Phase I of Project Halo, The Halo Pilot, which concluded in May 2003, was a six-month effort to determine whether state-of-the-art “knowledge representation and reasoning” technology is capable of producing computer applications that answer novel questions in Advance Placement (AP) chemistry – and also provide readable, domain-appropriate justifications for those answers. The project also identified two closely related challenges: (1) knowledge and question formulation requires highly specialized and expensive personnel (knowledge engineers), which pushes the development cost to about $10,000 per page; and (2) most of the evaluated system failures reflected insufficient expertise in AP Chemistry by the knowledge engineers creating the system’s knowledge modules.

Phase II
Halo Phase II will address these two issues directly by developing technology that will allow domain experts to formulate knowledge with decreasing dependence on knowledge engineers, and to pose questions and problems to the knowledge systems. Vulcan believes that achieving those goals will reduce the cost of knowledge formulation to levels comparable to textbook development, and will encourage scientists and educators to build an expanding body of machine-processable knowledge that will facilitate the Digital Aristotle’s role as a tutor and research assistant.

The 30-month Phase II effort will be undertaken in three stages: (1) a six-month design stage, (2) a 15-month implementation stage, and (3) a nine-month refinement stage. Three competing teams have been contracted by Vulcan, each with world-class skills and technology in five primary areas: knowledge representation and reasoning, knowledge acquisition, and intelligent interfaces, including natural language understanding, usability and system integration.

The teams are: Team SRI International, which includes team members from the University of Texas at Austin, Boeing Phantom Works, the University of Massachusetts Lowell and Kraka; Team Ontoprise, with team members from Carnegie Mellon University and the University of Colorado at Boulder; and Team ISX, with team members iSoco, Stanford Medical Informatics, the University of Southern California’s Information Science Institute, KSVentures and Klein Associates.

Two comprehensive user-driven evaluations, covering selected portions of the AP exams in biology, chemistry and physics (B), will be used to assess, guide and validate the research.

Future Work
Once a proof of concept for our knowledge formulation approach has been established, Project Halo will consider how knowledge modules might be integrated into interactive tutoring applications. We will also be looking into how knowledge modules might assist knowledge-driven discovery as part of the digital research assistant’s functionality.

Other potential research areas would try to understand how semantic Web and other technologies could help integrate two or more document-rooted knowledge modules to eventually produce an integrated scientific library capable of interdisciplinary and multi-resolution question-answering.