A Reflective Architecture for Knowledge Acquisition

The goal of the EXPECT project is to provide flexible tools for building and maintaining knowledge-base systems that are accesible to end-users. The key to this goal is EXPECT's reflective capabilities to analyze the contents of its knowledge bases and determine exactly what every piece of knowledge contributes to the task domain.

We have used EXPECT to build INSPECT, a decision-support tool to evaluate air campaign plans. As part of this work, we developed a structured grammar to represent planning objectives and goals in this domain. EXPECT has also been used in other domains, including transportation planning, protein synthesis, and configuration.

Research by EXPECT project members spans the areas of planning, knowledge acquisition, ontology modeling and construction, machine learning, knowledge representation, and natural language generation.

New areas of work include extending the EXPECT architecture to work with large-scale, high-performance knowledge bases , and to extend our work on air campaign planning.

The EXPECT project is funded under the DARPA/Rome Laboratory Planning Initiative (ARPI), DARPA's High-Performance Knowledge Bases program (HPKB), and DARPA's JFACC program.

The EXPECT Project

For a short introduction to EXPECT, click here.

A 3-minutes long Quicktime movie summarizes EXPECT. It was produced and directed by Bill Swartout. (The movie is a 1.7MB file.)

A recent overview talk about EXPECT and an overview of EXPECT's representation language.

EXPECT Project Members.

EXPECT is a project within the Intelligent Systems Division of the Information Sciences Institute of the University of Southern California.

Selected publications by members of the EXPECT project

For more information about EXPECT, send mail to expect@isi.edu.

Last updated: August, 1998.