Federation of Databases

1
Federation of Databases

• Archana Meka
• Ravikanth Kolli

2
OUTLINE

• Introduction
• Our Approach
• D2RQ Mapping
• Ontology Alignment
• Future Work
• Demo

3
INTRODUCTION

• Federated Database
• Transparently integrates multiple autonomous
  database systems into a single federation
• A component DBS can continue its local
  operations and participate in the federation

4
Ontology Driven Approach

• Ontology
• storage repository
• Container of commonly agreed upon knowledge
• Contains concepts and relationships between
  concepts

5
Languages

• OWL (Web Ontology Language)
• RDF (Resource Description Framework)
• N3 (Notation 3)

6
N3 (Notation 3)

• Readable language for data on the Web
• Compact and readable alternative to RDF's XML
  syntax
• Greater expressiveness

7
Example

• ltrdfRDF xmlns"http//www.atl.lmco.com/projects/o
  ntology/ontologies/weapons/WeaponsA2.n3"
• xmlnsa"http//www.atl.lmco.com/projects/onto
  logy/ontologies/weapons/WeaponsA2.n3"
• xmlnslog"http//www.w3.org/2000/10/swap/log
  "
• xmlnsowl"http//www.w3.org/2002/07/owl"
• xmlnsrdf"http//www.w3.org/1999/02/22-rdf-sy
  ntax-ns"
• xmlnsrdfs"http//www.w3.org/2000/01/rdf-sche
  ma"
• xmlnsxsd"http//www.w3.org/2000/10/XMLSchema
  "gt
• ltowlClass rdfabout"http//www.atl.lmco.com/
  projects/ontology/ontologies/weapons/WeaponsA2.n3
  ARMORED-
• COMBAT-VEHICLE"gt
• ltrdfssubClassOf rdfresource"http//www.
  atl.lmco.com/projects/ontology/ontologies/weapons/
  WeaponsA2.n3CONVENTIONAL-WEAPON"/gt
• lt/owlClassgt
• ltowlClass rdfabout"http//www.atl.lmco.com/
  projects/ontology/ontologies/weapons/WeaponsA2.n3
  CONVENTIONAL-WEAPON"gt
• ltrdfssubClassOf rdfresource"http//www.
  atl.lmco.com/projects/ontology/ontologies/weapons/
  WeaponsA2.n3WEAPON"/gt
• lt/owlClassgt
• ltowlClass rdfabout"http//www.atl.lmco.com/
  projects/ontology/ontologies/weapons/WeaponsA2.n3
  TANK-VEHICLE"gt
• ltrdfssubClassOf rdfresource"http//www.
  atl.lmco.com/projects/ontology/ontologies/weapons/
  WeaponsA2.n3ARMORED-COMBAT-VEHICLE"/gt
• lt/owlClassgt

8
Contd.

• _at_prefix xsd lthttp//www.w3.org/2000/10/XMLSch
  emagt .
• _at_prefix rdfs lthttp//www.w3.org/2000/01/rdf-sc
  hemagt .
• _at_prefix rdf lthttp//www.w3.org/1999/02/22-rdf
  -syntax-nsgt .
• _at_prefix a lthttp//www.atl.lmco.com/project
  s/ontology/ontologies/weapons/WeaponsA2.n3gt .
• _at_prefix owl lthttp//www.w3.org/2002/07/owlgt
  .
• aARMORED-COMBAT-VEHICLE
• a owlClass
• rdfssubClassOf aCONVENTIONAL-WEAPON .
• aTANK-VEHICLE
• a owlClass
• rdfssubClassOf aARMORED-COMBAT-VEHICLE .
• aCONVENTIONAL-WEAPON
• a owlClass
• rdfssubClassOf aWEAPON .

9
D2RQ Mapping

• Declarative language to describe mappings between
  relational database schemas and ontologies

10
Contd.

• Using D2RQ we can
• query a non-RDF database using SPARQL
• access information in a non-RDF database using
  the Jena Model API or the Sesame API
• do RDFS and OWL inferencing over the content of a
  non-RDF database using Jena's inference engine

11
Contd.
D2RQ V0.5 User Manual
12
Ontology Structure
D2RQ V0.5 User Manual
13
Cont.

• Ontology is mapped to database schema using
  ClassMaps and PropertyBridges
• ClassMap
• specifies how instances of the class are
  identified
• PropertyBridges
• specify how the properties of an instance are
  created

14
Ontology Alignment

• Problem
• given 2 taxonomies and their associated
  data instances, for each node in one taxonomy,
  find the most similar node in the other taxonomy,
  for a predefined similarity measure
• Ontology Alignment
• process of determining correspondences between
  concepts

15
Overview

• Exploit structural and lexical similarity
• Graph structure
• Node and edge labels
• Formulation within the iterative
  Expectation-Maximization (EM) scheme
• Suitable for taxonomies but can be used for
  edge-labeled ontologies using reification

May converge to local maxima
16
Contd.

• Map both ontologies using a many-one mapping
  method
• Considers schemas for matching between ontologies
• Use directed graph as underlying models for
  ontologies

17
EM Background

• Developed by Dempster, Laird and Rubin (1977)
• Maximum likelihood estimate of an underlying
  model from observed data (X) in the presence of
  missing values (Y)
• E-step
• Evaluate the likelihood of different models
  (Mn1) given a seed model (Mn)
• M-step
• Choose the best model and use it in the next
  iteration
• Generalized M-step
• Select a model that is better than the current one

18
Graph Matching Using GEM

• Treat the match assignments as the model
• Mixture model
• Given a data node, the correspondence with some
  model node is a hidden variable

19
E-Step

• Above equation is simplified considerably
• Involves finding the lexical similarity between
  
• the node labels
• We use the generalized M-step

20
Model Sampling

• Model space is large
• Random sampling from the model space
• Combine sampling with intuitive heuristics

Mn1
Mn1
Map-Parent Heuristic
Mn1
Mn
21
GEM Match
Recall 100 Precision 90
22
Partition Approach

• Breadth first traversal starting from the root
• Generate smaller ontologies
• Add fringe nodes for better matching of smaller
  ontologies
• Apply the EM to the smaller ontologies
• Generate the final Mixture Model with the nodes
  of both the ontologies

23
Contd.
24
Architecture
Ontology
RDBS
Ontology
SPARQL
D2R-Q
Results
User
Ontology Mappings
RDBS
Ontology
D2R-Q
25
Contd.

• Using D2RQ Mapping
• Extract the schema from the relational databases
• Generate an ontology with the schema
• Ontology Matching
• Generates the Correspondence between the concepts
  in both the ontologies

26
Contd.

• Querying
• Query a chosen ontology using SPARQL query
  language (currently implemented using Jena Graph
  API)
• Checks for similarity using the Correspondence
  Matrix.
• Returns all the results combined

27
Future Work

• Extending to multiple Information sources
• Developing a GUI for querying
• Extending the querying to conditions
• Providing global ontology with all the concepts
  from the local ontologies
• Enhancing the EM approach

28
References

• D2RQ V0.5 User Manual
• Federated database systems - Amith P.Sheth
• Inexact matching of ontology Graphs using
  expectation maximization - Prashant Doshi
• An Ontology Driven Approach to Data
  Integration - Agustina Buccella
• Observer System - E.Mena