This Small Business Innovation Research Phase I project will address the problem of automating complex multi-step tasks and negotiations for computer users. The focus is on developing representations and algorithms for a personal assistant agent, and on contributing the incorporation of these algorithms into negotiating meeting times. Personal assistant agents need effective models of their users' and others' preferences, which can be highly discontinuous in time and often complex. A key part of the research is on improving user preference learning algorithms, as well as designing a representation of the learned user model to facilitate interaction with the user.
Currently consumers waste significant time at work and at home interacting with applications to find information, conduct their work, connect with others, and generally organize their life. Personal assistant agents that automate tasks and negotiations based on their users' preferences, have the potential save people and businesses large amounts of time. By building a personal assistant agent platform Crono plans to focus on automating tedious multi-step tasks and interactions. The agent will, as it learns, completely remove the time burden of meetings scheduling from its users. The research promises to yield an enhanced understanding of how to learn about users, and how applications can interact with them and act on their behalf, building trust over time. If successful, the agents developed in this project will allow a new class of automated negotiation services to be fielded.
This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).
