Information Technology Reference
In-Depth Information
Communication Controller
The role of the CC agent is to facilitate communication between the different agent communities and
monitor any agent malfunction and inactivity. Changes to the foundation layer through newly added,
edited, or deleted local schemas are made known to the EDF by the CC agent. New data sources and
services offered by participating organizations are discovered by the CC based on UDDI registry que-
ries. Most multiagent development environments (like the JADE Project and the MadKit Project) are
sophisticated enough to recognize agents that join the community as well as to search and discover
other agents using some form of white- or yellow-page mechanism. The agent-monitoring and -facilita-
tor mechanism offers advanced functions such as tracking agents using unique names and addresses,
discovering services offered by the agents, and searching for agents providing specific services (JADE
Project, n.d.). The role of the CC is to capitalize on this core agent-monitoring operative and act as a
coordinator to manage the communication between the different IC agents and the EDF. The communi-
cation controller traces the addition or deletion of data sources from the FIS and tracks the status modes
(active, sleep, inactive, defunct) of all IC agents at the foundation layer. The communication controller
works in conjunction with the EDF to handle exceptions during subquery executions on a data source
and determine how results from multiple data-source subqueries can be filtered and combined before
the results are returned to the end user. The communication controller orchestrates this process through
service mediation and query mediation (JADE Project; MadKit Project, n.d.). Service mediation allows
agents to identify service components from the participating organizations and combine them to handle
specific client requests. After the service components are identified, the query mediator will carry out
the query execution on specific views or local schemas. Query mediation breaks down a user query into
subqueries against the local schemas. IC agents can execute the subqueries while being aware of the
schema restrictions, semantic differences, and structural limitations (Busse et al., 1999) by analyzing
the OWL-S document at the local data source.
Utility in the Running Example:
As the new stock investment firm decides to join the FIS for provid-
ing applicant information to the loan officer, the CC conducts service mediation to identify the URI that
describes all services offered by the investment firm. The URI reference is passed to the EDF, which
updates its OWL-S document with the new data source reference. The CC plays a vital role in ensuring
that the OWL-S document at the EDF stays in congruence with the data sources present in the foundation
layer. Query-mediation services allow the CC to designate and structure the subqueries for the newly
created IC agent and handle known exceptions in a manner that reduces the need for human interven-
tion. The results are returned to the CC, which then forwards this to the EDF for final verification and
compilation. This is shown in Table 1 (Row 4).
Interaction Controller
The architecture presented here involves the cooperative activity of multiple agent communities in con-
junction with ontology documents such as OWL-S that describe the semantic structure of the content
sources as well as the roles and responsibilities of the agent community. Figure 2 shows the multiagent
system that interacts with the data sources in an FIS. The IC agents can perform operations on the data
sources by drawing ontological inferences from semantic documents maintained at the data sources.
Semantic-based information abstraction has several real-world applications such as constructing sum-
mary indexes of available information from the data source, obtaining intensional repositories for data
Search WWH ::
Custom Search