The World Wide Web is the largest knowledge base that has ever existed. The Web was designed, however, to be understood by humans. The languages currently used to mark up Web pages do not provide the semantic annotation and inference capabilities that are needed to make Web information sources and other Internet entities (agents, devices, services, etc.) understood by computational processes.

The DARPA Agent Markup Language (DAML) will surpass current markup languages by representing semantic relations in machine-readable ways and enabling agents to reason about them. DAML is being developed as part of the Semantic Web initiative by teams of researchers at various companies and universities. Carnegie Mellon University is participating in DAML development through the Software Agents Group’s ATLAS project. ATLAS research objectives include the following:

1. DAML-based language for advertising and locating agents, ontologies, and other entities
2. Tools to facilitate the writing of advertisements and queries
3. Matching engines and agent infrastructure (e.g., ontology agents)
4. Ontologies for
   1. domain-specific agent services/capabilties
   2. describing service parameters and request preferences
   3. describing agent resources
   4. describing agent security parameters
   5. describing low-level services and platform characteristics
5. Agent search engines that serve as crawlers for the middle agent