Teaching Virtual Experts for Multi-Domain Collaborative Planning

Abstract

<p class="MsoNormal" style="text-align: left; margin: 0cm 0cm 0pt; layout-grid-mode: char;" align="left"><span class="text"><span style="font-family: ";Arial";,";sans-serif";; font-size: 8.5pt;">This paper presents an approach to rapid development of virtual planning experts that can collaborate to develop plans of action requiring expertise from multiple domains. The approach is implemented into a new type of software tool, called Disciple-VPT, which includes an extensible library of virtual planning experts from different domains. Teams of such virtual experts can be rapidly assembled from the library to generate complex plans of actions that require their joint expertise. The basic component of the Disciple-VPT tool is the Disciple-VE learning agent shell that can be taught directly by a subject matter expert how to plan, trough planning examples and explanations, in a way that is similar to how the expert would teach an apprentice. Copies of the Disciple-VE shells are used by experts in different domains to rapidly populate the library of virtual experts of Disciple-VPT.</span></span><span style="font-family: ";Arial";,";sans-serif";; font-size: 8.5pt;"></span></p

A Tool for Training and Assistance in Emergency Response Planning

“…Besides the development of the two reasoning trees, another result of this modeling process was the development of a modeling methodology which is presented in detail in [12]. Yet another result of this modeling process was the identification of the object concepts that need to be present in Disciple's ontology so that it can perform this type of reasoning.…”

“…The partially ordered set of these elementary actions represents the plan for performing the initial task. From such planning examples the agent learns general planning rules that allow it to plan the performance of similar tasks [12].…”