Learning through user instruction for personal assistants and agents

Main

Description

Status

Research

Publications

Demo

People

Funding

Links

Status

We have identified a set of interactions within a scenario where the user is purchasing a laptop computer that demonstrate how providing advice can customize the behaviour of the plan execution system in a deep way, modifying constraints and providing new action definitions.

For more details, see the Tailor project pages.

An initial set of sample interactions can be found here. They are intended for use with Spark, a PRS-based plan execution system being developed as part of Calo.

We are working to implement some of these interactions and provide an API to allow other modules within Calo to use the capabilities. We welcome comments on the approach.

<< Back to IKCAP