Using Contextual Knowledge to Resume Human-Agent Conversations when Programming the Intelligence of Smart Environments

This paper presents a hybrid technical solution towards addressing conversational interruptions when interacting (via a typed interface) with a virtual agent to program the intelligence of a smart environment. The AmI Solertis system is a ubiquitous programming environment that facilitates the definition of the behavior of a Smart Environment. To address the issue above, Ami Solertis introduces a mechanism that stores any unexpectedly interrupted conversations in a stack along with relevant contextual information. This context-sensitive information attached to the dialog, is used to re-establish a detailed context in the user’s working memory when resuming human-agent conversations, within.