Information Technology Reference
In-Depth Information
and specification. This provides the foundation technology, built upon an agreed and accepted standard
from W3C, for the capture, representation, exchange and storage of knowledge represented by business
rules and related data in XML format that can be potentially used and shared by software agents.
Recent initiatives to develop technologies for the “Semantic Web” (Berners Lee et al., 2001) make
the content of the Web unambiguously computer-interpretable, thus making it amenable to agent in-
teroperability and automatic reasoning techniques (McIllraith, et al., 2001). Two important technolo-
gies for developing Semantic Web are already in place - XML and the resource description framework
(RDF). The W3C developed the RDF as a standard for metadata to add a formal semantics to the Web,
defined on top of XML, to provide a data model and syntax convention for representing the semantics
of data in standardized interoperable manner (McIllraith, et al., 2001). The RDF working group also
developed RDF schema (RDFS), an object-oriented type system that can be effectively thought of as a
minimal ontology modeling language. Recently, there have been several efforts to build on RDF and
RDFS with more AI-inspired knowledge representation languages such as SHOE, DAML-ONT, OIL
and DAML+OIL (Fensel, 1997). While these initiatives are extremely promising for agent interoper-
ability and reasoning, they are at their early stages of development. In this chapter, we focus on the use
of more mature and widely used and available standardized technologies such as XML and DTDs to
represent knowledge. This approach, along with other initiatives, should allow researchers to develop
intelligent agent-based systems that are both practical and viable for providing intelligent decision sup-
port to users in a business environment.
xML and decision Trees for Knowledge r epresentation
The W3C XML specification allows for the creation of customized tags for content modeling. Customized
tags are used to create data-centric content models and rule-based content models. Data-centric content
models imply XML documents that have XML tags that contain data, for example from a database,
and can be parsed by application software for processing in distributed computing environments. XML
documents containing rule-based content models can be used for knowledge representation. XML tags
can be created to represent rules and corresponding parameters. Software agents can then parse and read
the rules in these XML documents for use in making intelligent decisions. Before making intelligent
decisions, the software agents should be able to codify or represent their knowledge. Decision Trees and
inductive learning algorithms such as ID3, C4.5 can be used by agents to develop the rule-based decision
tree. This learned decision tree can be converted into an XML document with the corresponding use
of a DTD. This XML document, containing the learned decision tree, forms the basis for knowledge
representation and sharing with other software agents in the community. We demonstrate architecture
for agent-based intelligent information systems to accomplish this.
xML and decision Trees for Knowledge r epresentation and exchange
Software agents for knowledge exchange and sharing in the agent community can exchange decision
trees represented in XML documents. For example, a new agent can learn from the knowledge of the
existing agents in the community by using the decision tree available in XML format in a knowledge
repository. The existence of this knowledge repository allows knowledge to be stored and retrieved as
needed basis by the agents and updated to reflect the new knowledge from various agents in the com-
munity. The explanatory power of decision trees from their ability to generate understandable rules and
Search WWH ::
Custom Search