from relational to semantics a methodology arka mukherjee, ph.d. founder / cto global ids david...
TRANSCRIPT
From Relational to Semantics
A Methodology
www.globalids.com
Arka Mukherjee, Ph.D.Founder / CTOGlobal IDs
www.revelytix.com
David SchaengoldDirector, Business SolutionsRevelytix
© 2012 Global IDs
2Proprietary
© 2012 Global IDs
3Proprietary
Objectives
1. The Bridge from Relational to Semantics
2. An Implementation Methodology
3. Review a Use Case
Introduction
© 2012 Global IDs
5Proprietary
Evolution of Data Representation
1970’s• Relational Models• Relational Databases
Optimized For:
• Speed• Volume
2010’s• Semantic Models• Graph Databases
Optimized For:
• Complexity• Integration
Improved Representations
© 2012 Global IDs
6Proprietary
Central Questions
Can software help us migrate from a relational-centric world to a semantic-based world?
Is there a clear path to providing better data integration and management of distributed data sets?
© 2012 Global IDs
7Proprietary
Partner Organizations
Focus:
- Data Profiling
- Data Mapping
- Enterprise Scale
Focus:
- Distributed Information Management
- Data Virtualization
- Data Federation
© 2012 Global IDs
8Proprietary
FocusRelational To RDF Mapping, Federation and Analytics
Methodology
© 2012 Global IDs
10Proprietary
Prototypical Customer
Pain Points:
Little Transparency or Traceability
Limited Analytical Capability
What is needed:
Quicker Time-to-Market
Reduced Cost of Integration
More Comprehensive Analytics
Access to More Data
Too Many Data Silos
High Costs of
Maintenance
© 2012 Global IDs
11Proprietary
2
Improve Quality
Methodology
3
Generate Portals
1
Create Transparency
5
Integrate / Federate
6
Emergent Analytics
4
Map to Ontology
6 Stages
© 2012 Global IDs
12Proprietary
Stage 1 : Create Transparency
1. Scan
2. Analyze
3. Organize
4. Create Maps
1. Convert Maps to Ontologies
© 2012 Global IDs
13Proprietary
Stage 2 : Improve Quality
Data Verification Data Validation
Data Stewardship Data Monitoring
© 2012 Global IDs
14Proprietary
Stage 3 : Generate Portals
For Any Data Type:
• Customers• Products• Employees• Suppliers• Vendors• Partners• SKUs• Sensors
© 2012 Global IDs
15Proprietary
Stage 4 : Map to Ontology
Map to:
• Company Ontology• Industry Ontology• Domain Ontology
An ontology is synonymous with a Business Vocabulary
© 2012 Global IDs
16Proprietary
Stage 5 : Integrate / Federate
• Populate Semantic Database
• Deploy SPARQL Query Processor
• Integrate Data, Enabling Federated Queries Over Distributed Data Stores
© 2012 Global IDs
17Proprietary
Stage 6 : Emergent Analytics
• Rapid Analytics
• Query any SPARQL End Point, Internally or Externally
© 2012 Global IDs
18Proprietary
Result: Lower Costs + Higher Automation
Financial Services Demonstration
© 2012 Global IDs
20Proprietary
Architecture
1. Scanning & Profiling
2. Export Mappings
3. Create Queries over Federated Sources
© 2012 Global IDs
21Proprietary
Starting Point
How is Data Defined?Where is Data Located?
© 2012 Global IDs
22Proprietary
Global IDs provides automated data discovery at enterprise scale
© 2012 Global IDs
23Proprietary
Revelytix provides enterprise datamanagement solutions using W3C semantic standards
A Distributed Information Management System is a layer above your current DBMS, similar to how a
RDBMS is a layer above a file system
• Both provide an additional level of abstraction
• Both bundle new computational capabilities into the system
• Both simplify the access to and use of data by applications and developers
© 2012 Global IDs
25Proprietary