Pleiades Project Home Page

Our objective is to create and demonstrate

machine learning methods that allow personal software agents to automatically customize to the needs of their users, and
methods for automated negotiation among these agents, in order to improve their effectiveness, robustness, scalability and maintainability.

In order to explore these issues, we have developed a collection of various software agents:

a Calendar Apprentice (CAP) which learns users' scheduling preferences (see our most recent paper, or click here for access to some of the data used in our experiments),
a Mosaic-based netnews reader (NewsWeeder) that learns users' reading interests ( try it!),
a Mosaic-based tour guide (WebWatcher) that helps you search for information, and learns from experience. See the project home page.
a Visitorhost which helps schedule visitors for technical briefings, and
a Personnel Information mediator which provides information about specific individuals and their jobs.

The Pleiades system diagram illustrates the interactions among these agents.

See our list of publications.

Research support: Arpa, Digital, Siemens

Project Members