Semantic Interoperability Symposium Stevens Institute of Technology Babio Hall, Room 202 April 11, 2007, 8:30-3:30 Participants Paul Albertelli, Lockheed Martin, Systems Engineer, [email protected]John Boardman, Stevens SSE, Distinguished Service Professor, [email protected]Michael DiMario, Lockheed Martin, Program Manager, [email protected]Rick Dove, Stevens SSE, Industry Professor, [email protected]Alec Gorod, Stevens SSE, Research Assistant, [email protected]Haibo He, Stevens ECE, Assistant professor, [email protected]Victor Lawrence Stevens ECE Dir Cntr for Intelligent Networked Systems victor lawrence@stevens edu Victor Lawrence, Stevens ECE, Dir Cntr for Intelligent Networked Systems, victor.lawrence@stevens.edu Roshanak Nilchiani, Stevens SSE, Research Assistant Professor, [email protected]David Nowicki, Stevens SSE, Research Professor, [email protected]JeanPiere Paillet, TopQuadrant, nc, [email protected]Art Pystor, Stevens SSE, Distinguished Research Professor, art.pystor@stevens.edu [email protected], attributed copies permitted 1 Brian Sauser, Stevens SSE, Assistant professor, [email protected]Yu-Dong Yao, Stevens ECE, Associate Professor, [email protected]
43
Embed
Semantic Interoperability Symposium - parshift.com · could make a major difference in semantic interoperability. ... Leadership Power ... Enterprise A Dimension n Factor m Enterprise
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Semantic Interoperability Symposium
Stevens Institute of TechnologyBabio Hall, Room 202April 11, 2007, 8:30-3:30
ParticipantsPaul Albertelli, Lockheed Martin, Systems Engineer, [email protected] Boardman, Stevens SSE, Distinguished Service Professor, [email protected] DiMario, Lockheed Martin, Program Manager, [email protected] Dove, Stevens SSE, Industry Professor, [email protected] Gorod, Stevens SSE, Research Assistant, [email protected] He, Stevens ECE, Assistant professor, [email protected] Lawrence Stevens ECE Dir Cntr for Intelligent Networked Systems victor lawrence@stevens eduVictor Lawrence, Stevens ECE, Dir Cntr for Intelligent Networked Systems, [email protected] Nilchiani, Stevens SSE, Research Assistant Professor, [email protected] Nowicki, Stevens SSE, Research Professor, [email protected] Paillet, TopQuadrant, nc, [email protected] Pystor, Stevens SSE, Distinguished Research Professor, [email protected]
Abstract – In the milieu of systems-of-systems, semantic interoperability is the frontier barrier of effective information exchange. Prime examples seeking solution: emergency responders cooperating in response management, war-fighters cooperating in real-time net-
t i t b i li ti i t ti i SOA d b icentric engagement, business applications interacting in SOA and web service environments. This symposium will present multiple profiles of high priority needs for semantic interoperability, then possible solution elements that would combine a new pattern processing/transformation technology with established semantic methodologies. Convergence of Interests: 1) A research thrust in the School of Systems and Enterprise isConvergence of Interests: 1) A research thrust in the School of Systems and Enterprise is developing knowledge, simulators and laboratory tools that will support research and live application in enterprise systems-of-systems investigations. 2) A research thrust in the EECS department is developing knowledge and devices for intelligent networks. 3) A new coprocessor technology is emerging that shows great promise as a semantic engine that p gy g g g p gcould make a major difference in semantic interoperability.Agenda:
• Introductions, Purpose, and ContextI t bilit d th EA R h P Ri k D• Interoperability and the EA Research Program, Rick Dove
• Interoperability and the Center for Intelligent Network Systems, Victor Lawrence• Semantic Interoperability in General, Jean-Pierre Paillet• Semantic Interoperability Challenges in DoD, Michael DimarioSemantic Interoperability Challenges in DoD, Michael Dimario• Semantic Interoperability Proposal Interests, Rick Dove• Semantic technology and employment, Jean-Pierre Paillet• Processor technology and employment, Rick Dove
■ Intelligent SoS research platform■ COD optimization (Configure-On-Demand)
■ Discovering compatibility knowledgeg p y g■ Discovering generating functions■ Discovering stability principles■ Web-mining self-learningg g■ Watching for SoS opportunities
Impact Some Research Possibilities
SoS Optimizing Virtual Environment
■ Optimal tactical war-asset deployment■ Optimal project partnerships■ Synergistic team construction
■ Principles for attractors & boundaries■ Principles for willful SoS stability ■ Principles for dynamic balance
Rational Systemic SecurityNew security embeds in enterprise DNA
■ Strategy assumes penetration■ Policy harmonizes with 7 realities ■ Security is systemic/embeddedy y■ Security becomes resilient■ Security practices initiative■ Cyber/Physical security is merged
$S it G
Old Newy y y g
■ Systems thinking is practiced
Impact
ROISecurity Gap
Key Pathfinder-Project ElementsImmediate benefits for strategy/policy ■ Compelling value propositionsAction is stimulated and guided■ Technology and knowledge roadmap
■ 20 participating organizations■ Federal co-sponsor■ 3 mos guided discovery workshops
■ Technology and knowledge roadmap
Lead-the-way role models are created■ Core participant organizations
From: The Reengineering Alternative - A Plan for Making your Current Culture Work, William E. Schneider. This is one of fivetables characterizing the core organizational cultures mapped by Schneider, a clinical psychologist who consults on organizationaldevelopment and change. The other four map the Structural, Power, Strategic, and Relationship focus factors. The vertical axisreflects what a culture pays attention to - the content - and ranges from what is to what might be. The horizontal axis reflects theprocess of how a culture makes decisions - and ranges from impersonal analysis to personal/interpersonal involvement.
Interoperability Profiling – Some Research Areas
Process WIP from Extended Enterprise SimulationProcessProcessSemanticsSecurity
Governance
WIP from Extended Enterprise SimulationNew focus – white papers and proposals
Next up
ProcessSemanticsSecurity
GovernanceGovernanceCultureEthicsSpeedSpeed
Innovation
???
Pool of interest, awaiting champions
???
What order should they come in? Implies importance weighting!Of the many factors in each dimension, which are relevant? When? What
th i i t f t i ht ?are their inter-factor weights?What is the nature of the compatibility function between two entities? Is it domain dependent? Is it objective (forcing-function) dependent?What is the nature of an optimization function?
Net Centric WarfareThe building of the tower of Babel by Pieter Bruegel, 1563 Oil on oak panel, Kunsthistorisches Museum Wien, Vienna
Tower of Babel
Many of the players speak a different language,Many of the players speak a different language, have different values, perceive things differently, accomplish things differently, and have different approaches to security….
This RFI seeks information on relevant technologies and concepts that could form the basis for a DARPA program to develop a Strategic Collaboration capability i e a network enabled
for a DARPA program to develop a Strategic Collaboration capability, i.e., a network-enabled collaborative environment (tools, processes, embedded metrics, organizing principles, new concepts of operations, runtime analysis, etc.) within which large numbers of people (1,000s of people from 100s of organizations, with different experiences, cultures, expertise, and concepts/constructs/languages) can come together in top-directed organizations, as well as p g g ) g p g ,emergent communities (communities of interest), in discourse and collaboration (each with their own language using their own tools, procedures, methods, etc.) to detect, recognize, conceptualize, represent, and experiment with the extremely complex problems of large scale SSTR/HADR operations, beyond any one person’s experience or comprehension to solve alone within the context of a single integrated environment (gaming simulation andsolve alone, within the context of a single integrated environment (gaming, simulation, and instrumented C2). Specifically, this RFI seeks information on relevant technologies, capabilities, and concepts in the following areas:
1. Semantic Glue: Provide individuals from diverse organizations and cultures, g ,speaking different languages and having different goals, the ability to collaborate in support of SSTR/HADR operations. “Semantic Glue” refers to the collection of technologies and capabilities that will enable strategic collaboration by providing lightweight, adaptive and interactive representation schemes (language, icons, and geospatial graphics) for
h i i f ti id f d i ith diff t f f t ( iexchanging information, across a wide array of devices with different form factors (ranging from cell phones to desk top computers), in environments with varying network bandwidth and reliability, while providing semantically meaningful information appropriate to the individual, organization, form factor, and network capacity.2 etc
Synopsis - Posted on Dec 15, 2005M difi ti t P i N ti P t d D t J 23 2007Modification to a Previous Notice Posted Date: Jan 23, 2007
SOLICITING WHITE PAPERS ONLYSOLICITING WHITE PAPERS ONLY. DO NOT SUBMIT A FORMAL PROPOSAL AT THIS TIME.
In FY07 this solicitation seeks innovative approaches and technologies forIn FY07, this solicitation seeks innovative approaches and technologies for achieving ontological mediation between different domains, vocabularies, information sources, and services. Capabilities of particular interest include ontological interoperability technologies which contribute to our ability to: (a) provide a coordinated view across multiple data sources and knowledgeprovide a coordinated view across multiple data sources and knowledge representations, (b) mediate over mappings between multiple data sources and formats into a format conducive to knowledge-based analytical tools and intelligence analysts, (c) consolidate, disambiguate and normalize multi-source data into concrete knowledge representations and (d) exchange and reusedata into concrete knowledge representations, and (d) exchange and reuse inference-capable knowledge representations across differing contexts.
In FY08, this solicitation seeks innovative approaches for information sharing and
JTRS and InteroperabilityPage 4/5 - GAO-06-955 Defense Acquisitions - Software-defined radios such as JTRS primarily use software rather than hardware to control how the radio works and becauseprimarily use software rather than hardware to control how the radio works and, because they are programmable, JTRS offers significant flexibility to meet a wide variety of needs. Rather than developing radios that are built to different standards and operate on different fixed frequencies, as was the case in the past, JTRS is to be a single, interoperable family of radios based on a common set of standards and applications. The radios are expected to not only satisfy the requirements common to the military’s three operational domains—air,not only satisfy the requirements common to the military s three operational domains air, sea, and ground—but be able to communicate directly with many of DOD’s existing tactical radios. To facilitate interoperability, JTRS will develop a set of waveforms (software radio applications) designed with the same operating characteristics as many of DOD’s existing radios.Depending on operational needs different waveforms could be loaded onto a JTRS radioDepending on operational needs, different waveforms could be loaded onto a JTRS radio and used to communicate with a variety of other radios.In addition to supporting interoperability, JTRS is to contribute to DOD’s goal of network centric warfare operations by introducing new wideband networking waveforms that dramatically increase the amount of data and speed at which the data can be transmitted. As y psuch, the waveforms would facilitate the use of maps, imagery, and video to support the decision making of tactical commanders at all echelons.Page 16 - GAO-06-955 Defense Acquisitions - Interim solutions for network interoperability: To achieve DOD’s desired networking capabilities, the waveforms must be able to interoperate reliably and securely with each other The optimal solution is to have thisinteroperate reliably and securely with each other. The optimal solution is to have this functionality performed inside JTRS radios as it reduces the overall footprint of the communication network. However, technologies and radio designs are not mature enough at this point to develop an interoperability capability that would function inside individual JTRS radios. Thus, for the initial increment, interoperability between the waveforms may be facilitated by developing gateway devices that reside outside of the JTRS radio This should
Real-time-programmable, general-purpose pattern recognition processorSemiconductor – RAM/ASIC/FPGAMassively parallel associative/systolic array with progress logicHybrid DFA/NFA machineFastest least expensive highest capacity pattern recognizerFastest, least expensive, highest capacity pattern recognizer
Constant speed, regardless of pattern complexityThroughput at wire speed, no matter how fast the wire gets (mostly)g p p , g ( y)
One processor chip can accommodate 1 million pattern featuresCan accommodate unlimited number of reference patternsS l f li it d f t / ttScales for unlimited features/patterns
No-penalty complexity breakthroughReduce false positives and negatives significantlyReduce false positives and negatives significantlyEnd the wire-speed race in network routing and filteringCollapses complex, exponential sequence alignment to a linear process Intelligence data-mining can be intelligent
(Kennen Technologies will make this technology available to appropriate Stevens research programs)
Semiconductor – RAM/ASIC/FPGAMassively parallel associative/systolic array with progress logicHybrid DFA/NFA machineFastest least expensive highest capacity pattern recognizerFastest, least expensive, highest capacity pattern recognizer
Constant speed, regardless of pattern complexityThroughput at wire speed, no matter how fast the wire gets (mostly)g p p , g ( y)
One processor chip can accommodate 1 million pattern featuresCan accommodate unlimited number of reference patternsS l f li it d f t / ttScales for unlimited features/patterns
No-penalty complexity breakthroughReduce false positives and negatives significantlyReduce false positives and negatives significantlyEnd the wire-speed race in network routing and filteringCollapses complex, exponential sequence alignment to a linear process Intelligence data-mining can be intelligent
This is a launch engine for the emerging semantic technology era
New Technology in Search of Application-Breakthrough Opportunities Overview of Kennen DHS Project
Patent-pending breakthrough co-processor IP for pattern recognitionA General purpose NFA pattern-processor chip – for data stream analysisIf fabricated as a single chip in current VLSI RAM technology:
Capacity: 1 million pattern features/chip, any number of patternsThroughput: ~ 1 GB/sec (per processor chip)Latency: One cycle, fixedProduction Cost: RAM equivalentPower/heat: Very lowScalability: Virtually unlimited
Can/will also be fabricated in ASIC/SoC and FPGA
A DHS Security Application in Search of a SolutionDHS S&T end-point hardware assisted security
SBIR Phase 1 Dec 2005 – June 2006 (6 mos)SBIR Phase 2 Mar 2007 – Sep 2008 (18 mos)
Processor CharacteristicsReal-time-programmable, general-purpose symbol-string processorSemiconductor – RAM/ASIC/FPGAMassively parallel associative/systolic array with progress logicHybrid DFA/NFA machineFastest, least expensive, highest capacity pattern recognizer , p , g p y p g
Constant speed, regardless of pattern complexityThroughput at wire speed, no matter how fast the wire gets (mostly)g p p , g ( y)
One processor chip can accommodate 1 million pattern featuresCan accommodate unlimited number of reference patternspScales for unlimited features/patterns
No-penalty complexity breakthroughNo penalty complexity breakthroughReduce false positives and negatives significantlyEnd the wire-speed race in network routing and filteringCollapses complex sequence alignment exponential to linear process
We have tried to contact you regarding our slashing of prcies. Special bundles now available. Please check us out for the chpest mdecationhttp://www.xxxLightCleaning.info
Visit our new online pharmcy store and saveup to 85% Todays special offers:Todays special offers: VIAG for as low as $1.62 P/D CIAL (super viag) for as low as $4.38 P/D LEVIT for as low as $4.44 P/D ... and much much more special offers today.
You Need 15 Minutes To Be Ready For Action. - All popular items are available - Free shipping worlwide
Eliminating False Positives and False negativesWith Additional ContextWith Additional Context
120 cells - 4 matches in the SPAM example
# Find spelling variations with additional context......# One entry from both of the following in either order# One entry from both of the following, in either order
# Viagra OR Levitra OR Ambien OR cheap OR meds OR pharmacy OR store# $n.nn | URL
Note: Combining the non-context and in-context examples takes 142 cells and yields 27 matches –this hit density could be used as a probability indicator.
Language Translation – Roughly Speaking(Spanish to English, for instance)
Preprocessing a document with 11,000 words -- W=11,000
1) Ignore non-essential words -- W=10,0002) Separate into overlapping 5-word phrases -- W=50 0002) Separate into overlapping 5-word phrases -- W=50,000
111 222 333 444 555222 333 444 555 666
333 444 555 666 777etc… etc …
3) Expand each word into a collection of 5 synonyms -- W=250,000
111a 222a 333a 444a 555a111b111c111d111e
222b222c222d222e
333b333c333d333e
444b444c444d444e
555b555c555d555e
+ + + +
4) Mark and remove duplicates -- W=240,0005) Pack and load -- W=120,000 … at 7 character/word average ≡ 840,000 Feature Cells
6) Score hits in 50-Gig reference stream at 1 Gig/sec = 50 secs, or 5 sec with 10x parallel 7) Assemble translated document – Note that total time will add pre and post processing
Showing Parallel Pattern Recognition Capabilityin the genomic-sequence problem space
I P l t t t #31 f d t t iti t k
data stream
Image: Processor emulator at step #31 of a data-stream recognition task
3 “regular expressions” defining 3 patterns,h ith 3 b tt t d b b d d “d ’t ”each with 3 sub-patterns separated by bounded “don’t care” gaps
Feature cells in the processor -- that implement the 3x3 pattern query shown above in the compile window
data stream character (“t”) satisfaction
PCI Card with FPGA and SDKOctober 2007
• DHS SBIR Phase 2 Prototyping Project (started March 2007)• Thrust: General purpose platform for data stream content analysis• Will produce application developer's platform: FPGA PCI board with SDK• Will produce application developer s platform: FPGA-PCI board with SDK• First boards available in October 2007• Boards will be used to seed market application development• Project focus will be security/network/NIC applications• Project focus will be security/network/NIC applications
...but other pattern-detection applications will also be seeded• Goal: at least 5 security application commercialization commitments
Goal: At least one ASIC/SoC development commitment• Goal: At least one ASIC/SoC development commitment• Goal: At least one NIC/LOM general purpose platform
Other Application Goals ( ith sec rit rele ant fall o t)• Other Application Goals (with security-relevant fall-out): ...Bio-sequence alignment, intelligence data mining, unstructured search, System-of-System semantic interoperability, semantic disambiguation, complex human-query unstructured search, etc
Application Developers Sought Explore benefits and nature of signature complexityExperiment with pattern program writingExperiment with re-casting security solutionsFeasibility and prototype-application developmentAd i f t ill tAdvise on features you will want
Enables new approaches to old problemsEnables innovative applications not previously economicEnables innovative applications not previously economic
Some applications may go to market on FPGA/PCI boardSome applications may want custom FPGA configurations Others will migrate to ASIC/SoC and VLSI processors
To Get InvolvedEmulator available nowEmulator available nowFPGA/PCI board with SDK available in October ... at costTraining classes will be providedCollaborative Web community will be developed
SummarySemantic Interoperability is a major and urgent problem. Major because large
ff t (JTRS SOA FCS) l l ti U t b jefforts (JTRS, SOA, FCS) rely on a solution. Urgent because major programs-in-process assume a solution will arrive in time.
Semantic Interoperability in-the-large does not have a demonstrable effectiveSemantic Interoperability in-the-large does not have a demonstrable effectivesolution. Current technical approaches impose restrictive and unrealistic constraints and controls on the scope of the problem (common “language”, customization expertise).
Interoperability in general is insufficiently respected/addressed by systems that must collaborate in a System of Systems (DoD Joint Programs).
Stevens sits at the nexus of advanced Systems Engineering education, Systems of Systems research, agile systems and systems in-the-large thinking, professional practicing system engineers, the interoperability-frontier
i h i h d d i dcommunity, the open-community research and education concept, and a breakthrough semantic processor technology.
The nature of the problem and the nature of these assets is a unique and
The nature of the problem and the nature of these assets is a unique and synergistic combination. New thinking, new technology, and new process will naturally converge on a reusable, scalable, effective solution.