The Tailor Project at ISI


Task learning by instruction in a procedural reasoning system

Description

Tailor allows end users to modify procedure definitions through short sentences describing the desired change. The procedure knowledge is represented in the SPARK language and shown to the user with a simple language generation interface.

More-->

Status

Our work on Tailor is part of the Calo project and began in May 2003.

We have recently integrated Tailor with SRI's IRIS platform and we are on track to deliver code for the Calo Year 2 test, in the Task Fulfillment Column.

More-->

Research

knowledge acquisition for procedural reasoning systems, search-based composition of compound expressions to match user text.

More -->

Publications

"An Analysis of Procedure Learning by Instruction", Jim Blythe. In Proceedings of the Twentieth National Conference on Artificial Intelligence (AAAI-05), 2005, Pittsburgh, Pennsylvania

"Task Learning by Instruction in Tailor", Jim Blythe. In Proceedings of Conference on Intelligent User Interfaces (IUI05). 2005, San Diego

"Incremental Formalization of Document Annotations through Ontology-based Paraphrasing", Jim Blythe, Yolanda Gil. In Proceedings of the International World Wide Web Conference (WWW-2004), 17-22 May, 2004, New York, NY, USA

More publications -->

Group Members:

DARPA's Personalized Assistant that Learns (PAL)