Extending Loosely Coupled Federated Information Systems Using Agent Technology - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 121

Figure 2. The extended FIS model

Personal Agent

Presentation Layer

Extended Directory
Facilitator Extensibility

Management Agent

Multi Agent System

Federation Layer
(uniform access language, schemantics,
agent architecture)

Communication
Controller

Interaction Controller
Agents

Data Sourc OWL

Financial Stability

Property
Value
Estimates

Employment
History

Credit Check

-Banks
-Investment Firms
-Insurance Companies

Foundation Layer
(data Sources for Loan application processing)

addressed by overlaying a multiagent architecture in conjunction with a set of ontological documents that
specify the semantic structure of the underlying data sources. The multiagent architecture can interpret
these ontological documents to discover changes and semiautomatically help with the extensibility of
the federated schema in the FIS by drawing semantic inferences from the ontological documents.
Figure 2 shows the OWL-S-based DL knowledge representation required for an agent system to interact
with the data source schemas. OWL-S provides the structured knowledge representation and maintains
abstraction from the underlying data models ( OWL-S , n.d.). OWL-S also provides the software agents
with a global knowledge of the federation's data model, including the subcomponents of the federated
schema and the service components offered by the different participating organizations.
The Semantic Web (Berners-Lee et al., 2001) portrays an architecture where agents can collect
content from diverse sources and exchange data enriched with semantics, facilitated by ontology docu-
ments. Agents can further assist in automated reasoning from the knowledge representations by apply-
ing inference rules on the structured collection of information represented in the ontology documents.
The model proposed in the next section mainly focuses on extending loosely coupled FISs using agent
technology.

An Agent-Based f ramework for Semantic extensibility

Semantic brokering models over dynamic heterogeneous data sources that use agents for subquery
processing and service discovery have been proposed in the past (Nodine, Bohrer, & Ngu, 1999). As