Document 199723

How to Implement Semantic
Solutions Without Blowing Up
Your Company
Paul Bradley - SVP, Account Management
1
2
DARPA Milestones
u 
Communications
ARPAnet (TCP/IP)
u 
Storage :
Relational Databases
u 
Describing
DAML-OIL
Ontology Web Language
Resource Description Framework
3
Data Management
u 
Create, Store, & Change
u 
Find
u 
Integrate
u 
Analyze
4
Business vs. Development
Functional
Requirements
Technical
Requirements
Coding &
System Test
UAT
Bug Fixes &
Enhancements
5
Big Data Paradigm Shift
6
Rules For The Next 40 Years
u 
Use - understood by machines
u 
Locate - descriptions are local
u 
Standards – interoperability
u 
Query - access distributed data
7
The Problem
Business people require access
to data they don't own
8
Why Are Semantic
Solutions The Answer?
Semantic-based technologies describe
enterprise things with precision
9
Why Semantics?
u 
Relate disparate data
- Same thing; different name
- Same name; different definition
u 
Uses abstraction to relate things
u 
Multiple layers of abstraction are needed
10
Thinking About Your Solution
u 
Use W3C open standards; no black boxes
u 
Deploy incrementally, no Big Bang
u 
Narrow your focus to show an early ROI
u 
Solve someone’s biggest riddle
u 
Be sure your solution is extensible
11
Pitfalls and Blind Alleys
Turns
u 
Believing ontologies are hard to build
u 
Minimizing the need for SME input
u 
Managing metadata with multiple tools
u 
Pushing existing BI solutions aside
u 
Attempting to replace data warehouses
12
Keep in Mind...
u 
u 
Data owners may build brick walls
Ontologies describe data in terms
understood by business people
u 
Minimal (or no) hardware requests
u 
Emphasize data stays where it is
13
And Above All....
Pick an Experienced Partner
14
Revelytix was founded in 2005 to
provide enterprises with a robust
analytical capability
Emergent Analytics
u 
Non-invasive
u 
Agile
u 
Low impact
15
How We Do It
u 
W3C standards for defining and describing things
u 
Tools to author / relate definitions and query data
u 
Create an “Enterprise Information Web”
u 
u 
Allow additional data elements / sources can be
added and new relationships defined at any time
Use a web-based collaboration tool allowing SMEs
and data modelers to work together in real time
16
We have a suite of products enabling you to
build an Enterprise Information Web
u 
Knoodl - web-based ontology development
and management
u 
Spyder - exposes any data structure as RDF
u 
Spinner - federated, parallel query
optimization engine
u 
Rex - rules validation engine
17
What We Can Do Now
SPARQL Queries &
Results
RDF Data Store
LOD Cloud
Planner
Indexes
Planner
Data Sources
Indexes
Optimizer
Cache
18
R2RML
Mapping
Optimizer
Cache
Exposes any data structure as RDF, accessible
using the SPARQL query language
u 
u 
Uses ontologies to map data schemas to standardized
RDF descriptions
Interacts with Spinner to federate data and optimize
query performance
u 
Exposes full power of SQL
u 
Reusable by reference
19
Federates a single query across
multiple SPARQL endpoints
Optimized description-driven, easily incorporated within
your existing IT infrastructure
u 
Supports complex analytics without a complicated
infrastructure
u 
Works with enterprise data where it is stored
u 
Delegates processing as close to the source database
as possible
u 
20
Rules Engine, executing rules written in
Rules Interchange Format
u 
Essential validation capabilities for results or data sets
exposed by Spinner
u 
Assures policy compliance
u 
Allows inferencing across data stores
21
Collaborative development tool and EIW registry
Most widely used web-based ontology tool
u 
u 
u 
Development and visualization:
u 
Ontologies
u 
Queries
u 
Dashboards
Repository:
u 
EIW ontologies
u 
Enterprise semantic metadata
22
Our Approach
u 
Determine gaps
u 
Identify quick hits
u 
Professional Services
u 
Organize communities of interest
u 
Software installation & configuration
23
Why Our Solution?
u 
Guarantees an immediate ROI
u 
Adds value to existing IT infrastructure
u 
Retains benefits from BI investments
u 
Easily implemented; no data is duplicated
u 
Quantifiable benefits
24
We Did It In New York
23
25
We re Doing It at the Pentagon
26
We Can Do It for You
u 
u 
Work with your data experts to create
ontologies
Configure our products to incorporate your
company’s data
u 
Deliver Emergent Analytics for you
u 
You see results in weeks, not years
u 
This stuff is real and we can prove it
27
Paul Bradley
September 27, 2011
[email protected]
28
29