Introspective predicates for explaining task execution in calo

From Semantic Portal Wiki

Edit

Reference:

1. Alyssa Glass, Deborah L. McGuinness. Introspective Predicates for Explaining Task Execution in CALO , Knowledge Systems, AI Laboratory, Stanford University (KSL-06-04), 2006

bibtex

@techreport { KSL-06-04 ,
  author = "Alyssa Glass, Deborah L. McGuinness", 
institution = "Knowledge Systems, AI Laboratory, Stanford University", 
number = "KSL-06-04", 
title = "Introspective Predicates for Explaining Task Execution in CALO", 
year = "2006", 
} 

abstract: In order to provide detailed explanations of the behavior of a cognitive agent, execution traces of the agent's behavior must be annotated with enough meta-data to support a wide range of behaviors. Particularly when answering complex questions, an explanation system must have access to information about many aspects of execution and planning. This document focuses on explanations for task-oriented processing within the CALO system. We plan to support a wide range of task explanations, providing CALO users with the ability to ask detailed questions about task execution and to engage in a mixed initiative dialogue with CALO about its past, current, and future task execution. We outline the needs that should be met by the Task Manager to allow usable, detailed explanations to be generated. These needs are organized around categories of "introspective predicates" - queryable facts that the Task Manager should support to provide necessary information to the explanation system.

download:

• paper: KSL-06-04.pdf
• slides: