Transcript DIA - Computação UFCG - Universidade Federal de Campina Grande

CIA 2003
7th International Workshop on Cooperative
Information Agents
DIA: Data Integration using Agents
Philip Medcraft, Ulrich Schiel, Cláudio Baptista
Universidade Federal de Campina Grande
Paraíba - Brazil
Introduction
Web  new requirements for data
integration.


The Semantic Web  ontologies to solve the
semantic heterogeneity.


Federated Database  Ontology.
Ontologies & software agents  data
integration
Introduction - DIA

A solution for semantic integration of data in a
federated database,
using mobile agents and ontologies.

Solve the following problems:



Unnecessary access to all data sources;
Excessive data flow.
Introduction



An ontology defines the concepts and
relationships between concepts of a
particular domain.
Ontology  the global schema of a
federation of DBs.
Rules  Global-Local Schema mapping
Mobile Agents

Some benefits:




Reduce network traffic;
Execute autonomously;
Adapt dynamically;
Plattform independent.
Design Patterns

Mobile agent patterns.

Itinerary pattern (modified).

Master-slave pattern.
DIA
DIA Architecture
DIA – Preparing the mobile agent itinerary

The Master-agent knows the schemas of the
federated databases.

Given a global query
 select the host whose schema attend the
query (Selected itinerary)
DIA – Preparing the mobile agent itinerary

Given the following query:
SELECT Cod, SUM(Credits)
FROM Customer
WHERE Cod = “02757”

A database whose schema does not contain
a corresponding attribute to the “Credits”
element, cannot attend the query.
DIA – Improving the Itinerary Pattern

Categories of queries:

visit all databases of the itinerary
 Static itinerary;

return before if query has been attended
 Dynamic itinerary.
DIA – Improving the Itinerary Pattern



Ex 1: “Give me the clients whose salaries are
above 1,000”. STATIC
Ex 2: “Is the salary of “Philip” above 1,000?”.
DYNAMIC
Ex 3: “Give me the sum of the salaries of
clients whose credit limits are above 1,000”.
STATIC

Ex 4: “Give me the name of a client where
the sum debits is superior to 10,000”.
DYNAMIC
DIA – Example
SELECT CPF, NAME, SUM(Credits)
FROM Customer
GROUP BY CPF
DIA – Local Integration

Concatenation

Integration (by an aggregate function)
DIA – Local Integration
Result example in XML syntax
DIA
Integrating
two XML
results
DIA – The federated system interface
The interface
DIA – Implementation and test




Java + JDBC
JXML (DAML-OIL  graphical representation)
Grasshopper
TEST with a federation of 3 DBMS:
Oracle + Interbase + SQL Server
Conclusion

Distributed corporations have
a group of well known databases which
maintain:


The knowledge of the stored information;
The stability of the federation members.
Conclusion

Ontology:



Global Schema;
Rules for mapping to local schemas
Mobile agents:


Reduce information flow;
Adequately choose the data sources to be
consulted.