From OntologPSMW

Jump to: navigation, search
[ ]

Contents

OntologySummit2015 Track C Session - Thu 2015-02-12     (1)

Session Co-chairs: MikeBennett     (1A)

This Track C session (Decision Making in Different Domains) will explore several approaches to automated inference in applications ranging from complex event processing and situation awareness to manufacturing.     (1C)

  • Date: Thursday, 12-Feb-2015     (1G1)
  • Start Time: 9:30am PST / 12:30pm EST / 6:30pm CEST / 5:30pm BST / 1730 UTC     (1G2)
  • Expected Call Duration: ~1.5 hours     (1G3)
  • Dial-in:     (1G4)
    • Phone (US): +1 (425) 440-5100 ... (long distance cost may apply)     (1G4A)
    • Skype: join.conference (i.e. make a skype call to the contact with skypeID="join.conference") ... (generally free-of-charge, when connecting from your computer ... ref.)     (1G4B)
      • when prompted enter Conference ID: 843758#     (1G4B1)
      • Unfamiliar with how to do this on Skype? ...     (1G4B2)
        • Add the contact "join.conference" to your skype contact list first. To participate in the teleconference, make a skype call to "join.conference", then open the dial pad (see platform-specific instructions below) and enter the Conference ID: 843758# when prompted.     (1G4B2A)
      • Can't find Skype Dial pad? ...     (1G4B3)
        • for Windows Skype users: Can't find Skype Dial pad? ... it's under the "Call" dropdown menu as "Show Dial pad"     (1G4B3A)
        • for Linux Skype users: please note that the dial-pad is only available on v4.1 (or later; or on the earlier Skype versions 2.x,) if the dialpad button is not shown in the call window you need to press the "d" hotkey to enable it. ... (ref.)     (1G4B3B)
    • instructions: once you got access to the page, click on the "settings" button, and identify yourself (by modifying the Name field from "anonymous" to your real name, like "JaneDoe").     (1G5A)
    • You can indicate that you want to ask a question verbally by clicking on the "hand" button, and wait for the moderator to call on you; or, type and send your question into the chat window at the bottom of the screen.     (1G5B)
    • thanks to the soaphub.org folks, one can now use a jabber/xmpp client (e.g. gtalk) to join this chatroom. Just add the room as a buddy - (in our case here) summit_20150212@soaphub.org ... Handy for mobile devices!     (1G5C)
  • Discussions and Q & A:     (1G6)
    • Nominally, when a presentation is in progress, the moderator will mute everyone, except for the speaker.     (1G6A)
    • To un-mute, press "*7" ... To mute, press "*6" (please mute your phone, especially if you are in a noisy surrounding, or if you are introducing noise, echoes, etc. into the conference line.)     (1G6B)
    • we will usually save all questions and discussions till after all presentations are through. You are encouraged to jot down questions onto the chat-area in the mean time (that way, they get documented; and you might even get some answers in the interim, through the chat.)     (1G6C)
    • During the Q&A / discussion segment (when everyone is muted), If you want to speak or have questions or remarks to make, please raise your hand (virtually) by clicking on the "hand button" (lower right) on the chat session page. You may speak when acknowledged by the session moderator (again, press "*7" on your phone to un-mute). Test your voice and introduce yourself first before proceeding with your remarks, please. (Please remember to click on the "hand button" again (to lower your hand) and press "*6" on your phone to mute yourself after you are done speaking.)     (1G6D)
  • Please review our Virtual Session Tips and Ground Rules - see: VirtualSpeakerSessionTips     (1G7)
  • RSVP to gruninger@mie.utoronto.ca with your affiliation appreciated, ... or simply just by adding yourself to the "Expected Attendee" list below (if you are a member of the community already.)     (1G8)
  • Please note that this session may be recorded, and if so, the audio archive is expected to be made available as open content, along with the proceedings of the call to our community membership and the public at-large under our prevailing open IPR policy.     (1G10)

Attendees     (1H)

Chat Transcript     (1I)

[09:14] Mike Bennett: This Track C session (Decision Making in Different Domains) will explore several approaches to automated inference in applications ranging from complex event processing and situation awareness to manufacturing. (1C) Agenda (1D) Ontology Based Information Centric Tactical Edge Networking, Joseph Kopena (Bellerophon Mobile) (1D1) From Semantic Complex Event Processing to and Ubiquitous Pragmatic Web 4.0, Adrian Paschke (Freie Universitaet Berlin) (1D2) Process Ontologis for Smart Objects in Manufacturing, Michael Gruninger (University of Toronto) (1D3) Situation Awareness and Decision Making, Ken Baclawski (Northeastern University) (1D4)     (1I1)

[09:30] sjBreiner: Hello everyone. I (sjBreiner) am a new researcher at NIST. Nice to meet you.     (1I2)

[09:30] Michael Grüninger: @sjBreiner: Welcome!     (1I3)

[09:31] Mark Underwood1: Greetings all - Downloading of slides posted on the Track page. . the first set is athttp://ontolog.cim3.net/file/work/OntologySummit20> -Centric-Tactical-Edge-Networking--JosephKopena_20150212.pdf     (1I4)

[09:32] Mark Underwood1: The page for the session is http://ontolog-02.cim3.net/wiki/ConferenceCall_2015_02_12     (1I5)

[09:32] Mark Underwood1: @SJBreiner - @NISTGov very welcome here     (1I6)

[09:39] Michael Grüninger: Ontology Based Information Centric Tactical Edge Networking, JosephKopena (Bellerophon Mobile)     (1I7)

[09:39] Mark Underwood1: Live tweeting at Tweetchat - Hashtag #ontologysummit     (1I8)

[09:39] Mike Bennett: == Joe Kopena presenting ==     (1I9)

[09:53] Adrian Paschke: What is the relation between the ICN paradigma and peer-to-peer principles as used e.g. in publish subscribe systems?     (1I10)

[09:55] Mike Bennett: "Platform Framework" questioner: you typed your question next to the hand, which made it think the question is your name. Please can you toggle the hand and type your question in the wide box below. Thanks.     (1I11)

[09:56] Amanda Vizedom: re: slide 6 and relevance determination. This harkens back to work done on DARPA Command Post of the Future - Entire program was focused on getting the relevant/important info (real-time, deployed context) into the decision-maker's awareness without overwhelming. The program ended up focusing mostly on viz, but the part I worked was a proof-of-concept ontology-based, context-sensitive monitor of incoming data.     (1I12)

[09:56] Peter P. Yim: @JosephKopena - what does the ICN acronym stand for ("Information Centric Network"?)     (1I13)

[09:59] Mark Underwood1: @Amanda CPOF - <sentimental smile >     (1I14)

[09:59] Conrad: Are the SPARQL queries distributed across the network layer that is collecting data & Meta-data?     (1I15)

[09:59] Amanda Vizedom: @JosephKopena - Were you aware of the DARPA/CPoF work as background? If so, can you comment on how the #IoT framework and technologies (or other changes since) change the possibilities and challenges here?     (1I16)

[10:00] Dennis Wisnosky: was it necessary to use a rules engine. If yes, what?     (1I17)

[10:00] Adrian Paschke: @JoeKopena: How is the trade-off between reasoning complexity and real-time processing addressed in an ICN network?     (1I18)

[10:00] Tara Athan: A fire is an observation? This sounds like conflation of an observation with the subject of the observation.     (1I19)

[10:01] Amanda Vizedom: I lost track of which slide # we're on - someone clue me back in?     (1I20)

[10:02] Peter P. Yim: slide 10 (? not totally sure)     (1I21)

[10:08] Adrian Paschke: currently on slide 12     (1I23)

[10:08] Mark Underwood1: Ref - UCore (I wrote the Wikipedia entry for this) - small world     (1I24)

[10:10] Michael Grüninger: now on slide 13     (1I25)

[10:10] Dennis Wisnosky: What is the concept lattice??     (1I26)

[10:11] Terry Longstreth: @Tara - in military doctrine, "fire" refers (usually)to artillery. I think that's what's meant here. See: http://armypubs.army.mil/doctrine/DR_pubs/dr_a/pdf/atp3_09x24.pdf     (1I27)

[10:11] Michael Grüninger: Slide 14 now     (1I28)

[10:12] Mark Underwood1: Q for @TJKopena Curious how mil-caliber network security overlay was / would be implemented; this a significant challenge for real-world military KR R&D     (1I29)

[10:13] Mark Underwood1: Esp for distributed / mobile soldier scenario     (1I30)

[10:14] Dennis Wisnosky: How many FTE's were involved in this and how long did it take?     (1I31)

[10:15] Bobbin Teegarden: If SPARQL and RDF not quite right tools, what would be a better set?     (1I32)

[10:17] Dennis Wisnosky: From the former DoD CTO and CA for Business - VERY IMPRESSIVE!     (1I33)

[10:17] Steve Ray: Lost sound?     (1I34)

[10:17] sjBreiner: Next slide set?     (1I35)

[10:18] Steve Ray: Sorry, just me. Will dial in again.     (1I36)

[10:18] Mike Bennett: == Adrian Paschke presenting ==     (1I37)

[10:19] sjBreiner: Thanks     (1I39)

[10:20] Joe Kopena: @Mark: Security is addressed at multiple levels. The whole system can ride over standard mil link, network, and transport security in exchanging data. CBMEN also includes a key-policy attribute-based encryption component to enable really fine grained control of who can read what content. The challenge is that securing the metadata while also permitting the kind of reasoning we're doing is well beyond the current state of the art. So that part basically defaults to being secured by the underlying network, and organized into tiers. E.g., only a certain level of authorized node might be able to decrypt the metadata I'm generating. So, everybody might be able to participate in caching and forwarding content, without being able to read it, but not everyone can actually match.     (1I41)

[10:22] Joe Kopena: @Dennis: Thanks! Our component, constructing the reasoner, ontology, and autogen, basically had 3 FTEs. The larger CBMEN system as a whole had a number of teams, mostly with good sized groups of their own (we were by far the smallest). The project as a whole ran for about 2 years, much of it absorbed in live field testing.     (1I43)

[10:23] Mark Underwood1: @JoeKopena - Did you have work in a secure skiff etc.?     (1I44)

[10:23] Joe Kopena: @Bobbin: I am not sure exactly what would be best. I have some thoughts on annotated SPARQL, but in part my interest in participating here and reengaging with the academic community in particular is to refresh myself on what alternatives are out there.     (1I45)

[10:24] Joe Kopena: @Mark: No, all of the work is unclassified. The only parts necessarily done in secure facilities (outside the field tests) were testing with actual tactical radios.     (1I46)

[10:25] Joe Kopena: A big part of that is we used commercial smartphones as the computational platforms     (1I47)

[10:27] Mark Underwood1: @JoeKopena - Ack'd . Security issues tended to limit diffusion of KR work I've worked on     (1I48)

[10:28] Amanda Vizedom: @JoeKopena - Did you work with mixed human sensor output (e.g., SPOTREP, SITREP), non-human sensor output (e.g., UGS), and/or mixed (e.g., JSTARS report)? If so, did you notice or suspect any differences in information characteristics or metadata required?     (1I49)

[10:30] Joe Kopena: @Dennis: The concept lattice is basically the structure created by organizing the concepts and how they relate to each other. I use "lattice" specifically because we started by identifying traits of the concepts and formally constructing a mathematical lattice, though I admit that quickly broke down in favor of more intuitive/informal organizing.     (1I50)

[10:30] sjBreiner: @AdrianPaschke (slide 14): I would have expected sequencing to be associative. Is there a difference between (A;B);C and A;(B;C)?     (1I51)

[10:31] Joe Kopena: @Amanda: Yep, definitely aware of CPoF, though I was not aware of efforts incorporating ontologies into it. We've previously done work integrating CPoF to mobile networks and handheld situational awareness apps.     (1I52)

[10:34] Joe Kopena: @Conrad: In the current system it's the reverse, metadata is shared throughout and queries are conducted locally. We've done some work in the past on aggregating metadata and queries for scaling across larger networks, but it wasn't really needed here yet and didn't make it into the final system (the end-program demo networks had 100 nodes). The one bit of optimization that does happen is the network is softly partitioned into communities based on network connectivity, physical location, and force organization, which bounds the spread of content/metadata until the communities merge.     (1I53)

[10:35] Joe Kopena: @Dennis: Yes, there is a basic rule network system implemented to execute some of the OWL semantics and to proactively match subscriptions. All of that was written by scratch, essentially constructing elaborate structures within the SQLite database underneath.     (1I54)

[10:35] Amanda Vizedom: @JoeKopena - As a DARPA program, CPOF became more viz focused as low-hanging fruit there became obvious, and that is where the bulk of the experimentation went. The transitioned and now-existent CPOF system includes only that aspect of what was done during the DARPA program. However, GITI, Cycorp, and a few designated SMEs worked on prototype Battlespace Monitor, using Cyc (I was PM/PI for the Cycorp part).     (1I55)

[10:37] Joe Kopena: @Adrian: Lots of similarity between P2P and ICN. I personally see the latter as essentially either a generalization or a lower-layer version of the former, depending on how you look at it.     (1I56)

[10:38] Evan Wallace: CIM - Computation Independent Model in MDA     (1I57)

[10:39] Joe Kopena: A major tradeoff we made between reasoning and performance here is in only using OWL-Lite for the ontology, and the reasoner being essentially solely focused on ABox reasoning (the final version actually did no TBox reasoning, it's just necessary in this application, though it could and previously did incorporate some). A big point here is that network and physical connectivity delays tend to be substantial, mostly dwarfing this level of processing. The approach is also geared toward infrequent, ad hoc content rather than routing high frequency streams. Though it's also very applicable to discovering the latter.     (1I58)

[10:42] Terry Longstreth: Sorry, I have to leave.     (1I59)

[10:45] Amanda Vizedom: @JoeKopena - I believe, based on experiences like the CPOF one and others in very different domains, that there is a pragmatically-definable language-and-reasoner system, beyond DL in expressivity but requiring only selected (or selectable/tunable) features of something like CyCL, CL, or IKL. IME many people have built special purpose reasoners that could be so described, but we still lack something like this in the public domain. I see such a reasoner as having enormous enabling potential. Given your comments about the mismatch of the CBMEN challenge with RDF/SPARQL, do you agree? Any further comment?     (1I60)

[10:48] Amanda Vizedom: @AdrianPaschke - slide 41, pragmatic web. Do you see the markup of actions, offers, etc., via schema.org, for example, as a step in this direction?     (1I61)

[10:52] Joe Kopena: @Amanda: Definitely agree. I don't have a super great handle on what exactly that language should be, but it's probably some combo of explicit taxonomy + cardinality & range/domain restrictions + basic rules + some mechanism for incorporating dynamic data (e.g., "current position").     (1I62)

[11:06] Mike Bennett: == Ken Baclawski presenting ==     (1I64)

[11:06] Adrian Paschke: Michael: do you also address the case that RFID events are uncertain?     (1I65)

[11:07] Mike Bennett: This content can be regarded as an introduction to the material covered in this Track C.     (1I66)

[11:07] Michael Grüninger: @Adrian: uncertainty is handled by nondeterminism, but we do not handle probabilities     (1I67)

[11:11] Joe Kopena: @MichaelG: Once you identify problematic axioms through adding to the minimal ontology, is there particular methodology or even a tool you're using to develop lemmas boosting the proof search, like the transitivity example you have on slide 10, or are you coming to those more intuitively?     (1I68)

[11:12] Amanda Vizedom: @MichaelGruninger: Thank you for calling out the identification of axiom types by both usefulness to, and consequences for, particular kinds of reasoning tasks. IME some folks who have worked only with DLs do not realize the potential and feasibility of this. Have you written up, anywhere, more about what you all did along these lines?     (1I69)

[11:12] Michael Grüninger: @JoeKopena: We don't have a rigorous methodology yet -- we are still trying to generalize the intuitions that lead to the lemmas.     (1I70)

[11:15] Michael Grüninger: @AmandaVizedom: The first appearance of this work was M. Grninger, S. Shapiro, M. S. Fox, and H. Weppner. Combining RFID with Ontologies to Create Smart Objects. International Journal of Production Research, 48(9):2633-2654, 2010. Our more recent work has focused more on the theorem proving heuristics, but this has not been published yet     (1I71)

[11:18] Adrian Paschke: @sjBreiner: yes (A;B);C means C must sequentially follow the complex event interval [A,B], whereas A;(B;C) means the complex event interval (B,C) sequentially follows A. In a timepoint semantics this leads to counter intuitive detections, e.g. an event instance sequence a, b, c would match to the event pattern A; (B;C), but also an event instance sequence "b, a, c", which is not the expected result.     (1I72)

[11:18] Mark Underwood1: @Adrian -what is your email address?     (1I73)

[11:19] Adrian Paschke: paschke@inf.fu-berlin.de     (1I74)

[11:20] Mark Underwood1: @MichaelG Is there a short answer to "How did you find the PSL as applied to production orders?" (I tried to use it awhile back)     (1I75)

[11:21] Adrian Paschke: @KenBaclawski: Did you work together with Dieter Galwick from Oracle?     (1I76)

[11:21] Dennis Wisnosky: John Boyd was the father of the F16.     (1I77)

[11:22] Michael Grüninger: @Mark Underwood: We translated the PSL axioms, the manufacturing ontology extension, and object history axioms into the syntaxes of Vampire and Prover9, and gave them directly to the different theorem provers. As I (hurriedly) mentioned in my presentations, we were able to answer most queries quickly.     (1I78)

[11:22] Dennis Wisnosky: Congrats to Mike and Mike on an excellent session of practical projects today.     (1I79)

[11:22] Michael Grüninger: Although some queries were stubbornly slow (which is why I also presented the different heuristics at the end)     (1I80)

[11:21] Adrian Paschke: I have been working with Dieter for over 20 years, going back to when we were working with Jim Gray.     (1I81)

[11:25] Mark Underwood1: @MichaelG - so PSL did serve as a reasonable KR framework. Interesting. (I have some familiarity with SAP PO data structures - it's complex, involves different sorts of knowl for discrete (cars) vs. process (drugs) manufacturing))     (1I82)

[11:26] Peter P. Yim: ^^ "Dieter Gawlick"     (1I83)

[11:27] Michael Grüninger: @MarkUnderwood: Yes, PSL was a good framework. A lot of time was spent just trying to understand the intended semantics for the many different classes of process plans within the domain. PSL can represent them, but the question is exactly what kinds of constraints we are supposed to represent in the domain     (1I84)

[11:29] Marcela Vegetti: @MichaelGrninger     (1I85)

[11:22]. Have you done the translation manually? Is there any tool to do that?     (1I86)

[11:30] Mike Bennett: @Ken thanks for the shout-out to FIBO     (1I87)

[11:31] Michael Grüninger: @MarcelaVegetti: we have a translator for Common Logic to Prover9 that Torsten Hahmann has written     (1I88)

[11:32] Peter P. Yim: Great session, great talks! ... a Big THANK YOU to the organizers and the speakers for today's session!     (1I89)

[11:33] Michael Grüninger: Next week's session: Synthesis I and Communique Outline     (1I90)