W3C WWW2004: Using the
W3C Standard OWL in
Adaptive Business Solutions
Updated: MAY-2004
© Network Inference 2004
Today’s Agenda

Network Inference and the Semantic Web

Semantic Web Business Case

Core Adaptive Enterprise Use Cases:

WWW2004
New York

Business Inferencing

Semantic Data Integration
Adaptive Enterprise Software Solutions

Fortune 500, Financial Chart of Accounts Management

NATO Country, Battlespace Awareness Desktop

Startup, Healthcare Patient Care System

Top 5 Reasons Why OWL Matters

Top 5 Reasons Why Description Logics Matter
© Network Inference 2004
NI and the Semantic Web


Network Inference is a Semantic Web innovator:

Prof. Ian Horrocks is a key contributor to OWL

Dr. Deborah McGuiness is a key contributor to OWL

Rob Shearer is a key member of the RDF Data Access WG

Network Inference, in 2003, unveils the first commercial OWL
inference platform and successfully deploys with several
customers.
Network Inference is committed to building Adaptive
Enterprise software solutions using Semantic Web
specifications and technologies

WWW2004
New York
The greatest business impact, and adoption potential, will be
to assist companies by lowering costs, and to improve
adaptive capabilities of traditional enterprise deployments.
© Network Inference 2004
The Business Case

Who cares that it is “Semantic Web?”



WWW2004
New York
Usually not our customers…
From a business point of view:

It can drastically lower operational costs

It provides powerful adaptive (new) business capabilities

It enables automation of business activity (standardized)
Because the underlying technology can:

Eliminate proprietary, non-interoperable metadata

Enable Machine interpretability of semantics (vs. syntax)

Create “reasonable” metadata about architecture layers
© Network Inference 2004
Use Case: Business Inferencing

What is it, and why should I care?

Business Inferencing is machine visibility into
operational data, semantics, and business rules


WWW2004
New York
Previously, any comparable capabilities were via highly
proprietary metadata markup embedded inside tools
Business Inferencing enables dynamic
applications to reason with and reclassify
corporate data

Thus enabling machine access to business knowledge –
automated use of all data and rules – instance data too.

It is used as a platform for application development

Replaces the business rules tier and manages business
vocabularies at the infrastructure level – saves $$$
© Network Inference 2004
Use Case: Semantic Data Integration

What’s different, why can’t the established
vendors simply add-in these capabilities?

Semantic Data Integration is the use of ontology
as a mediating vocabulary for disparate
underlying sources – a virtual hub and spoke


Most vendors are committed to their data
architectures, OWL is best used in the “core” –
not as an “add-on” to an existing COTS product.

WWW2004
New York
Unlike previous “business object” or “bus” style approaches,
ontologies are conceptual languages at a higher abstraction –
they don’t have to map 1:1 with underlying systems
© Network Inference 2004
Full automation will not come “for free” with simple plug-ins,
however, dramatic improvements are achievable
Use Case: Semantic Web Services*

Why is “meaning” important in web
services, SOA, and grid computing?

Avoid transformation code between data sets

Unambiguously capture service profiles

Enable dynamic discovery of services


Enable dynamic collaboration of services


Use reasoners to infer service descriptions and capabilities
Enable rich, automatic, service orchestration

WWW2004
New York
Use reasoners to locate services in “yellow pages”
© Network Inference 2004
Process layer will be far more automated with semantics
* Not a current customer deployment from NI
Fortune 500 Customer

Business Problem: Costly, untimely reporting of sales
in a chart of accounts

Solution: OWL-driven adaptive platform for the allocation
of unit sales and application of automated business rules
Market
Segments
Financial Analysts
WWW2004
New York
© Network Inference 2004
Business
Inference
Platform
Product
Classifications
NATO Country Customer

Business Problem: Inflexible IT systems prohibit robust
visibility to changing battlespace IT systems

Solution: Easy XQuery access (with built in class and
instance level inference) to intelligence data from disparate
sources – enabling visibility into rapidly changing data,
classifications, and rules.
OWL
RDF
XQuery
WWW2004
New York
Business Data
Inference Quality
© Network Inference 2004
M
e
d
i
a
t
i
o
n
Web
Services
Operational
Systems
Intelligence
Databases
Healthcare Startup Customer

Business Problem: Costly adaptations to patient
knowledge base with rapidly changing classifications

Solution: Business inferencing solution automatically
reclassifies complex knowledge structures on-the-fly
Web Portal
X-Query Interface
Nurses,
Doctors
Patient Families
WWW2004
New York
© Network Inference 2004
A-Box
T-Box
(inference)
(inference)
Symptoms & Resources
Why Description Logics?

Consistency – query results, across vendor
implementations and instances, should be consistent

Performance – Although performance metrics
depend on model constructs, OWL-DL supports
highly optimized inference algorithms

Predictable – semantics are mathematically
decidable within the model, reasoning is finite

Foundational – provides a baseline inside
applications for layered semantic models

WWW2004
New York
Reliability – if the answer to a query is implied by
any of the model data, it will be found – guaranteed.
© Network Inference 2004
Top 5 Reasons for OWL

Loose-coupling – semantics may be decoupled
from the application code (or parsing algorithms)

Machine-actionable – automated decisions can
be made from interpretable inferences

Highly expressive – can capture core elements
of EER, UML, and frame-based systems

Precision – language checking available to
prevent inconsistent/contradictory model semantics

Fun acronym! – OWL is named for the owl in
Winnie the Pooh, who spelled his name WOL
WWW2004
New York
© Network Inference 2004
WWW2004, New York
With the W3C’s standard OWL, everybody
can finally enable truly adaptive, standard
application architectures.
[email protected]
© Network Inference 2004
WWW2004
Backup Slides:
why owl matters to IT systems…?
© Network Inference 2004
Why OWL Matters – Reason #5

Semantics are loosely-coupled


WWW2004
New York
Characteristic

OWL ontologies are schema representations,
independent of application code and RDF models

OWL markup is easily stored and referenced in a
loosely-coupled registry/repository style architecture
Benefits

Semantics are late-bound, thereby supporting an
evolutionary – not static – network model for changing
data meanings and business rules

Semantics may be easily federated in simple markup

Semantics may be loosely-coupled to instance data
© Network Inference 2004
Why OWL Matters – Reason #4

Semantics are machine-actionable


WWW2004
New York
Characteristic

OWL is syntax (not graphical) grounded in XML & RDF

OWL uses consistent, standard schema semantics

Supports well-scoped classes, properties (class
relationships), instances and instance relationships
Benefits

Parsers, modelers, reasoners, and transformers are
available today

DL guarantees 100% decidability and computational
completeness – any two DL reasoners should come up
with the same (all possible) answers to queries
© Network Inference 2004
Why OWL Matters – Reason #3

OWL is more expressive


WWW2004
New York
Characteristic

Rich set of built-in simple properties, property
characteristics and restrictions

Not just hierarchical or taxonomic (like most XML)

Not just two-dimensional (like ER/RDBMS)

Allowable, functional, multiple inheritance
Benefit

More closely models “real-world”

Axioms may be used to model rules directly into the
model (compare with OCL-type approaches)
© Network Inference 2004
Why OWL Matters – Reason #2

OWL is more precise

Characteristic

Relationships are atomic and unambiguous


WWW2004
New York
Unlike UML/ER/XML, properties have stand-alone meaning

Disallows over-riding attributes (no semantic ambiguity)

DL enforces consistency

Within a context, semantics can be 100% unambiguous
Benefit

Reasoners can accommodate uncertain/unknown data

Both explicit and implicit facts are available via a
mediated query capability
© Network Inference 2004
Why OWL Matters – Reason #1

OWL is a FUN acronym (and apt!)


Characteristic

OWL = wisdom

OWL is named for the owl in
Winnie the Pooh (who spelled his
own name “WOL”)
Benefit

WWW2004
New York
© Network Inference 2004
Makes people
smile and laugh!
The End.
[email protected]
© Network Inference 2004
Descargar

W3C WWW2004 - World Wide Web Consortium