The%20Challenge%20of%20Data%20Interoperability%20from%20an%20Operational%20Perspective

1
The Challenge of Data Interoperability from an
Operational Perspective

• Workshop on Information Integration
• Todd Hughes
• DARPA/IXO

2
The Challenge of Data Interoperability

• Different weapons systems employ different
  transformation algorithms, which can induce
  degradation of accuracy due to rounding and
  approximations . . . Americas warriors are not
  fighting with a common positional picture,
  despite technological advances. (JP 3-09.3)
• Different Services, and even different weapons
  platforms within the same Service, use a variety
  of coordinate formats. A working knowledge of
  different formats is often lacking between
  Services, which may induce error and/or delays to
  mission accomplishment. (Gruetzmacher et al
  2002)
• In addition to the major combat units, OEF/OIF 4
  required specific capabilities supplemented
  within the theater . . . As these additional
  capabilities were added, the tracking of
  decomposed unit level organizations throughout
  the process became unmanageable . . . Although
  most of this information is available somewhere,
  it cannot be easily discovered or accessed in a
  timely manner, and does not render itself for
  easy manipulation by computers. (Chamberlain et
  al 2005)

3
Data Interoperability In Illustration

• Data interoperability is rightly regarded as a
  pervasive, longstanding, and costly problem
• Why has data interoperability research not
  enjoyed support commensurate with the severity of
  the problem?
• It may help to consider data interoperability
  technology from a general operational perspective

Full Interoperability

100
Serviceable Interoperability
Non-interoperability
DataOperational Readiness
0
Operational Timeline ?
4
Data Interoperability Conventional

• Data management systems do work well at the
  enterprise level
• Certain aspects make them not a good fit for
  agile, dynamic organizations . . . such as the
  military in times of conflict
• In times of conflict, operational timelines are
  to far too short for the enterprise acquisition
  model

100
Update
Update
Deployment
DataOperational Readiness
Conventional interoperability approaches take too
long to achieve operational readiness and require
too much downtime
0
Operational Timeline ?
5
Data Interoperability Semantic Web

• The Semantic Web offers a better value
  proposition by enabling interoperability on an
  open scale
• Price of admission is still high Semantic Web
  services, ontology engineering, client
  applications, service oriented architectures,
  etc.
• Downstream benefit of new semantic capabilities
  are difficult to quantify

100
DataOperational Readiness
The Semantic Web promises greater extensibility
and robustness, but the startup costs are still
enormous.
0
Operational Timeline ?
6
Data Interoperability An Alternative

• Future joint and multinational military
  operations will need to integrate in days, not
  months
• Data interoperability technology must support ad
  hoc communities of interest with their respective
  legacy data sources
• If the technology enabling the integration is
  fully assured, a serviceable level of integration
  may be sufficient
• What technology framework would make this
  possible?

100
DataOperational Readiness
An Alternative Approach Achieve serviceable
interoperability rapidly Evolve toward full
interoperability Learn to update more efficiently
the next time
0
Operational Timeline ?
7
Proposal A Data Translation Appliance

• Delivers transparent interoperability
• Intercepts content from a data sources or their
  applications
• Transforms content into syntactically well-formed
  and semantically equivalent messages
• Sends messages to other sources or their
  applications
• Field configurable by data users, not database
  administrators
• Onboard intelligence, learning, and interface
  capabilities
• Enables users to encode information
  transformation routines in a timely manner
• Modular domain knowledge
• Pluggable Community of Interest Ontologies
• Supports operational integrated information
  exploitation
• High data throughput
• Ruggedizable
• Certifiable
• Forward-deployable

• Semantic Technology Embedded in Translation
  Device
• Data indexing and taxonomization
• Data model extraction and enrichment
• Data model alignment
• Data translation and transformation

8
Prospects for a Data Translation Appliance

• Appliance model of integration is not without
  challenges, both technical and operational
• However, it may be a useful framework for
  discussion about the elements a successful
  research program in this area
• Technical barriers
• Metrics
• Evaluation scenarios
• Deployment strategies
• Operational Impact
• Return on investment
• Such a device would provide a tangible means for
  data users to be directly involved in meeting
  their operational objectives