Programming Research Group

Hybrid Declarative Framework for Multi Agent Systems (HDFMAS)

HDFMAS deals with issues faced by developers in agent architectures and agent languages, as introduced by Wooldridge and Jennings in "Agents Theories, Architectures, and Languages: A Survey". In heterogeneous distributed environments, information may be stored in different formats, using diverse synonymic variable names to represent one and the same entity, e.g. the representation of a person's attribute as "Lastname" or "Surname" in WSDL, in tables of databases or in data objects of some programming language.

The paper"A strategic framework for multi-agent cooperative dialogue" states that information exchange is one of the key interoperability problems. In an agent dialog, even if all the agents understand a common declarative language, computation of a query is difficult to handle due to heterogeneity of information in backend systems. The cooperative agents engaged in heterogeneous environments should be capable of exchanging and sharing complex information, their intentions, plans and strategies among them.

As already figured out in the paper "Intentional Programming for Agent Communication", an Agent Communication Language (ACL) must be declarative and have at least a small number of primitives that are necessary to facilitate agent interoperability. Therefore, HDFMAS is being developed by following Gruber's translation approach and a Probabilistic Object Model (POM) for data representation to make the agent communication independent of any declarative language and to enhance the collaborative work among software agents.

Contact: Mandeep Singh

Address

Universität Tübingen

Wilhelm-Schickard-Institut für Informatik

Jasminweg 21

72076 Tübingen

GERMANY