From OntologPSMW

Jump to: navigation, search
[ ]

Contents

OntologySummit2015 Track D Session - Thu 2015-02-26     (1)

Session Co-chairs: MarkUnderwood     (1A)

  • Abstract
    The existence of standards ��� both official and de facto ��� can dramatically influence the software development life cycle for ontology projects. This is especially for greenfield efforts, which can peg existing vocabulary, interoperability settings, test harnesses and verification processes to new projects. Standards may be essential for domain-specific data quality assurance. Standards also have a sociotechnical purpose. Communities of Interest (CoI) behind a standards effort can supersede a standard by concentrating expertise and collecting artifacts related to the standard. Because the world of ���things��� is by definition vast, standards can facilitate connecting software to devices by offering abstractions that impact domain-specific knowledge of the devices. This is helpful for building ontologies. That said, software development is a deregulated engineering process, and many successful software ventures have succeeded by ignoring or incorporating bits and pieces of unacknowledged standards work ��� sometimes creating new de facto standards in the process. The purpose of this track is to help potential IoT ontology developers understand the standards landscape ��� both official and de facto.     (1B)

Approach     (1C)

  • Survey "official" IoT standards     (1D)
  • Identify de facto standards     (1E)
  • Identify related standards, projects, bodies not officially designed as IoT or WoT     (1F)
  • Highlight domains where ontology efforts:     (1G)
  • Discuss Challenges: Power mgmt., security, signal post-processing, provenance, signal quality, discovery, metadata, network issues, Big Data     (1H)
  • Related work: Modsim (e.g., Yang Song, et al. 2012, Sensor Fusion, Linked Open Data, augmented reality, Software Defined Networks (SDN)     (1I)
  • Lessons from history: middleware, intelligent agents, CEP, embedded systems, DoD fusion     (1J)
  • Retrospective: Related Lessons from Ontology Big Data 2014     (1K)

Session 2 - Participant Q and A    (1L1)

  • Q1. What does an ontology "sprint" designed to integrate with an IoT standard look like?     (1L1A)
  • Q2. You've been asked to add surveillance video to an existing IoT project that only incorporates ground sensors designed for a municipal parking system.     (1L1B)
    • (a) What features in a video stream standard would facilitate integration?     (1L1B1)
    • (b) How should the original parking system ontology have been designed to facilitate this later upgrade?     (1L1B2)
  • Q3. Gruninger's presentation demonstrated a project that relied upon the Process Specification Language (ISO 18629). Given that most ERP systems have extensive (ERD?) data models but do not "implement" the standard, what approach would you recommend to a team that wishes to integrate a new generation of factory floor sensors for a discrete manufacturer?     (1L1C)
  • Q4. Miller's presentation discussed how the Constrained Application Protocol (CoAP, RFC 7252) can be used to web-enable low level devices like switches and valves. Does this suggest a "layer" for ontologies?     (1L1D)
  • Q5. What design patterns seem to be "curated" by standards organizations (using language from Track C "Engineering Bottlenecks" conversation in the 2014 Summit)?     (1L1E)
  • Q6. Should standards organizations consider sub- or micro-ontologies to accompany their taxonomies? E.g., NIST working groups often produce taxonomies to support documents like the Cloud Security Standards and the Big Data Public Working Group, but not ontologies. Should that work be left to others, or does the absence of ontology ingredients handicap wider adoption?     (1L1F)
  • Q7. One of the concerns expressed about standards like AllJoyn, originally started by Qualcomm's Connected Experiences group, is that it's not vendor-neutral. To what extent can an ontology standard mitigate or simply inherit these problems?     (1L1G)
  • Q8. If one or more standards set forth alerting mechanisms, such as the alerting in OGC SensorML 2.0 and the Oasis Common Alerting Protocol used in NIEM and elsewhere, how can ontologies address the problem of alert fatigue? Are there cross-cutting principles from KR that can be applied to guide user interfaces, Big Data variety, information aggregation / concentrators and outputs from predictive analytics?     (1L1H)
  • Q9. A topic touched upon (Hodges, "semantic workflow") but not fully addressed is standardization of workflow and orchestration. In a Big Data, M2M, IoT, sensor-rich world powered by DevOps, what role should an IoT ontology play in marshalling, monitoring, managing IoT resources -- perhaps including (as Ram Sriram suggests) roles performed by human agents? Should ontologies be designed for IoT subsystems following the ontology-oriented design pattern for BPEL suggested by Nitzsche et al (2007) and Aslam et al (2006)?     (1L1I)
  • Q10. Should ontology efforts be focused on "meta" or "global" problems, or more modest, predictable challenges? For instance, an ontology that models the behavior of a class of sensors and cultivates developer attention to device reliability, error recovery and resilience might have greater value than something more ambitious. (Consider the previous summit commentaries about uncertainty.)     (1L1J)
  • Q11. What should the role of an ontology (and an ontologist) be when a new generation of sensor devices is introduced -- resulting in not only a mix of sensor streams, but a mix of decision processes, data models and predictability?     (1L1K)
  • Q12. Can a "hard-coded" system (one in which knowledge representation elements are embedded by a developer oblivious to ontologies), be "Ontology Light"-enabled by connecting to various levels or connectors specified by an IoT standard?     (1L1L)
  • Q13. How is an ontology for decentralized (edge-aggregated or preprocessed streams) different from a centralized one? Should the ontology itself be distributed to the edge? How does this affect software engineering for distributed nodes?     (1L1M)
  • Q14. How can the use of IoT ontologies enhance device provenance (Compton, Corsar, Taylor 2014) through standardized audit, query, metadata or configuration management? Given short device life cycle for smart devices (e.g., smart phones, wearables), what needs to be in place for ontologies to co-evolve?     (1L1N)
  • Q15. The ontoCAT integration with R (Kurbatova et al., 2011) for bioinformatics and the rOntorion R package suggest possible collaborations for IoT analytics, perhaps machine learning through R. Microsoft hosted a 2014 conference on machine learning in which a Microsoft blog post argued that "some of the most exciting work being done to reap value from the Internet of Things (IoT) involves taking data insights to the next level using machine learning (ML)." What role should ontologies play in this arena?     (1L1O)
  • Q16. Would the presence of a readily usable ontology for a class of IoT devices facilitate white hat / black hat IoT attacks, penetration testing?     (1L1P)
  • Q17. Are there principles from the 2008 Ontology Summit Metadata for Ontologies discussion that should be reinvigorated for IoT settings?     (1L1Q)

Prepared presentation material     (1M)

Prepared presentation material (slides) can be accessed by clicking on each of these links:     (1M1)

Additional Resources     (1N)

  • Dial-in:     (1O5)
    • Phone (US): +1 (425) 440-5100 ... (long distance cost may apply)     (1O5A)
    • 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.)     (1O5B)
      • when prompted enter Conference ID: 843758#     (1O5B1)
      • Unfamiliar with how to do this on Skype? ...     (1O5B2)
        • 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.     (1O5B2A)
      • Can't find Skype Dial pad? ...     (1O5B3)
        • for Windows Skype users: Can't find Skype Dial pad? ... it's under the "Call" dropdown menu as "Show Dial pad"     (1O5B3A)
        • 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.)     (1O5B3B)
    • 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").     (1O6A)
    • 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.     (1O6B)
    • 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_20150226@soaphub.org ... Handy for mobile devices!     (1O6C)
  • Discussions and Q & A:     (1O7)
    • Nominally, when a presentation is in progress, the moderator will mute everyone, except for the speaker.     (1O7A)
    • 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.)     (1O7B)
    • 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.)     (1O7C)
    • 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.)     (1O7D)
  • 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.     (1O11)

Attendees     (1P)


Chat Transcript     (1Q)

[09:24] Mark Underwood: Questions for our Q&A session today are on the session page: http://ontolog-02.cim3.net/wiki/ConferenceCall_2015_02_26 (refresh for latest Qs)     (1Q1)

[09:30] Ravi Sharma: is there presentation material link for download?     (1Q2)

[09:32] Mark Underwood: @Ravi, the questions for discussion are inline on the session page     (1Q3)

[09:34] Amanda Vizedom: May I suggest muting all now, to lessen background noise on recording?     (1Q4)

[09:38] Amanda Vizedom: Two clarification questions re Q1: (1) Are you asking about an ontology sprint generally, or specifically and IOT standard integration sprint? (2) Are you asking about strictly-agile sprints only, or agile-influenced sprints?     (1Q5)

[09:39] Liana Kiff: My reaction is that the ontology is no different than any other piece of code,     (1Q6)

[09:40] Mark Underwood: Michael: ambiguity detection / mapping incompleteness     (1Q7)

[09:41] John Morris: The question of an ontology sprint says to me "proof of concept" -- and that means all the things you need for a POC.     (1Q8)

[09:41] John Morris: Such as "sponsor", "business need", "decision criteria" etc.     (1Q9)

[09:42] John Morris: Projects fail typically not for technical reasons, but because there isn't senior exec sponsorship and real business need. OK "business pain" as much as "opportunity".     (1Q10)

[09:44] Liana Kiff: Ontologies are equally likely be over-engineered as other elements of a software architecture, and Agile engineering can help to focus the teams on the most critical end-to-end features the ontology must support.     (1Q11)

[09:46] Ravi Sharma: Is the short agile sprint to facilitate integration only or is it to develop that part of overall ontology of IOT?     (1Q12)

[09:49] Bobbin Teegarden: The problem with SCRUM/sprints and agile is that the methodology works off 'requirements', and those end up usually trying to fix a current problem in the current environment, and do not incorporate thinking about what 'could be', or making BIG changes in current systems. If the IoT is really a paradigm change, the 'requirements' would entail systemic vision of big changes. Doesn't that beg a more system architecture approach to systemic change?     (1Q13)

[09:49] Mark Underwood: @Michael stds may have implicit ontologies     (1Q14)

[09:50] John Morris: Interesting counterpoint to @Liana - and with acknowledgement that there is often over-engineering in various domains -- but agile is often underengineered -- consider the question of "data modeling", which is often done badly and even disparaged. Data modeling is a close as most get to any kind of systematic conceptual model, which ultimately is an ontology. So "ontology could be considered the ideal complement to the typical challenges of agile". With ontology, agile will have a sound foundation.     (1Q15)

[09:50] Amanda Vizedom: Summarizing what I said on the call: (Defining sprint loosely, and assuming the availability of test harness, device for testing, etc., as Mark posited) my answer is that an #IOT ontology standard integration sprint looks very much like a ontology sprint in other contexts in which development is focused on building/improving a device or system with specific functioning... (cnt'd)     (1Q16)

[09:51] Mark Underwood: @adrian - LIfe cycle mgmt is a major issue on smaller devices, can build a less expressive ontology     (1Q17)

[09:51] Michael Grüninger: If we want to design an ontology that is correct wrt to a given IoT standard, what are the IoT standards that would have the highest impact in terms of providing it with an ontology?     (1Q18)

[09:51] Ravi Sharma: I would think it would be as analogy to databases (ODBC JDBC) a more universal interface such as for SCADA and DAS because even for wearable medical devices, there might be multiplexing and datastream concept to support installed or future installable devices.     (1Q19)

[09:52] Adrian Paschke: from my point of view the support for life cycle management is important. In particular, automated extraction of submodules, automated testing of the submodule within the constraints of the devices, automated download from repositories and deployment into the IoT sensors / devices     (1Q20)

[09:52] Peter P. Yim: ^^ == Q1. What does an ontology "sprint" designed to integrate with an IoT standard look like?     (1Q21)

[09:52] Peter P. Yim: Mark Underwood: == Q2. You've been asked to add surveillance video to an existing IoT project that only incorporates ground sensors designed for a municipal parking system. (a) What features in a video stream standard would facilitate integration? (b) How should the original parking system ontology have been designed to facilitate this later upgrade?     (1Q22)

[09:54] Evan Wallace: @Liana: In my experience, ontologies are likely to be either over-engineered or under-engineered, and probably more often over.     (1Q23)

[09:55] Amanda Vizedom: ...(continuing from [09:50]) Specifically, it would start with definition and clarification of the capabilities to be developed, sufficiently so that tests can be written for them. Then it would move into focused ontology development, making whatever additions, changes, mappings, etc., are needed to integrate the standard, where that is understood operationally as "to make the system pass those tests". The availability of frequent testing gives continual feedback during the sprint, and all on the team understand that at the end of the sprint, just that much should be accomplished if the sprint is to be a success... (ctnd)     (1Q24)

[09:55] Ravi Sharma: Also Mark it would help if not only format but metadata about the video are also incorporated in the stream?     (1Q25)

[09:56] Adrian Paschke: for fine grained semantic annotation of video streams, identifiable temporal and spatial media fragments are needed     (1Q26)

[09:56] Matthew West: On Q2 you have to invoke the "No Magic" principle. If you did not plan for something, there is no reason why you should be able to add it later.     (1Q27)

[09:58] Peter P. Yim: Mark Underwood: == Q3. Gruninger's presentation demonstrated a project that relied upon the Process Specification Language (ISO 18629). Given that most ERP systems have extensive (ERD?) data models but do not "implement" the standard, what approach would you recommend to a team that wishes to integrate a new generation of factory floor sensors for a discrete manufacturer? ... ref. http://ontolog-02.cim3.net/w/index.php?title=ConferenceCall_2015_02_26&oldid=15492#hid1H     (1Q28)

[09:58] Ravi Sharma: Q3 ERP to IoT one of aspects would be to interject CRM or user experience history as part of increasing usage of IOT in ERP     (1Q29)

[09:59] Amanda Vizedom: ... (cntd from [09:55]) Also, as Mark commented or elicited after my first comment: There is likely to be a welcome need for cross-functional interaction here. The ontology team will be better integrated into the overall engineering team than is often the case. And at the end of the sprint, system test might address whether the device can now communicate what it senses in such a way that any standard-compliant (or standard-aware) consuming system can correctly interpret it.     (1Q30)

[10:00] Ram D. Sriram: Since we are talking about humans also being integrated into the system we should talk about "social sensors," which could be patterns from social interactions. For example, consider the chat from the Boston Marathon bombers incident. "Social sensors" could have recognized certain patterns that could have foreseen this event. Also, photos are an important part and linking these in a standard manner would be useful. Ramesh Jain was going to talk about this on March 12, but he has requested that his talk be rescheduled.     (1Q31)

[10:01] Ravi Sharma: Q3 another aspect would be links to not only ERD but ERD data models in the context of Big Data     (1Q32)

[10:01] Liana Kiff: @Evan over-engineering is even more likely if the ontology isn't being exercised for a real-world use case. So the deep integration and co-development with the solution, is likely to make both the ontology and the solution more effective.     (1Q33)

[10:01] Adrian Paschke: the question here is, if the ontology is addressing the operational execution level of ERP, such as workflow execution, or the tactical and strategic decision layer of enterprise resource planning.     (1Q34)

[10:01] Michael Grüninger: Q3: The need here is for bridge ontologies between the ERP concepts and the IoT concepts     (1Q35)

[10:03] Mark Underwood: @Michael - Interop may be the design pattern best recognized by IoT integrators     (1Q36)

[10:04] Mark Underwood: Q4. Miller's presentation discussed how the Constrained Application Protocol (CoAP, RFC 7252) can be used to web-enable low level devices like switches and valves. Does this suggest a "layer" for ontologies?     (1Q37)

[10:04] Amanda Vizedom: @Bobbin [09:49] - I'm speaking from an agile-influenced, but not strictly agile perspective. With that caveat, I think you're right that the architectural and forward-looking perspectives are important. But it's actually extremely useful for such big goals to be divided into smaller bits, for which corresponding milestones and sprints can be defined. IME everyone is happier, and the results are better, when this can be accomplished. It's very effective at mitigating the tendency to spend a year or more working before rubber and road meet, at which point resources and will might not be available for major course corrections.     (1Q38)

[10:04] Ravi Sharma: these sensors fill the repository very fast.     (1Q39)

[10:05] Peter P. Yim: Mark Underwood: == Q4. Miller's presentation discussed how the Constrained Application Protocol (CoAP, RFC 7252) can be used to web-enable low level devices like switches and valves. Does this suggest a "layer" for ontologies? ... ref. http://ontolog-02.cim3.net/w/index.php?title=ConferenceCall_2015_02_26&oldid=15493#hid1L1D     (1Q40)

[10:05] Michael Grüninger: Q4: Is this an issue of granularity or abstraction?     (1Q41)

[10:06] Ravi Sharma: or architecture layer? in terms such as middleware?     (1Q42)

[10:07] Christopher Spottiswoode: Matthew, is that layers or modules?     (1Q43)

[10:08] Liana Kiff: Our experience so far is that a layered approach is desirable. Separating protocols from meaning, for example. S     (1Q44)

[10:09] Peter P. Yim: [OT] @MarkUnderwoood - I updated your session page and took out all html tagging in your question section ... (in general, please avoid html tagging, and leave the wiki text in pure text, so the system can automatically generate Purple Number tagging for the content.)     (1Q45)

[10:09] Amanda Vizedom: @LianaKiff and @EvanWallace -- I strongly agree that over-engineering is a frequent problem and exacerbated when there is a lack of real-world use case and corresponding tests. I'd add that under-engineering and just flat out bad engineering are similarly a problem and simliarly mitigated by applying practices such as requirements-based, testable milestones, and frequent testing.     (1Q46)

[10:10] Ram D. Sriram: @Mathew: Time is another important aspect of IoT. Events are key to IoT. We should be thinking about event ontologies and related standards.     (1Q47)

[10:10] Bobbin Teegarden: @Amanda Yes, nicely put. As long as the vision of what could be is big (or creative) enough, and as long as the vision is adjusted and the 'bits' are tuned, including widening based on emergent patterns. As Matthew West said above (paraphrase), if you don't plan, you might not be able to make adptations and BIG changes later based on what you're seeing emerge.     (1Q48)

[10:11] Sudarsan Rachuri: I would like to put few points. 1) How to filter signal from noise from all the devices including humans as devices 2) How do we understand the uncertainty both in measurement and also in the model of the system, 3) How do we incorporate "learning" and Bayesian learning for ontology     (1Q49)

[10:12] Christopher Spottiswoode: Yes Matthew, you are right, Does it matter? But the underlying assumption is still a component architecture facilitating reuse but recognizing granularities and precisions.     (1Q50)

[10:12] Mark Underwood: Q5. What design patterns seem to be "curated" by standards organizations (using language from Track C "Engineering Bottlenecks" conversation in the 2014 Summit)?     (1Q51)

[10:12] Ravi Sharma: We have two thoughts going in parallel - ontology the embeds different IoT or ontology tagged to one device or sensorgroup? any pro-cons?     (1Q52)

[10:12] Adrian Paschke: yes, typically indirect event-based communication     (1Q53)

[10:14] Ravi Sharma: @Adrian - in what context?     (1Q54)

[10:15] Michael Grüninger: Q5: What are the lists and/or working groups that we can send this question to?     (1Q55)

[10:15] Amanda Vizedom: I'm not sure I understand Q5, really.     (1Q56)

[10:16] Mark Underwood: Q6. Should standards organizations consider sub- or micro-ontologies to accompany their taxonomies? E.g., NIST working groups often produce taxonomies to support documents like the Cloud Security Standards and the Big Data Public Working Group, but not ontologies. Should that work be left to others, or does the absence of ontology ingredients handicap wider adoption?     (1Q57)

[10:16] Ravi Sharma: I had in previous Summits mentioned the need for developing standards ontology that would make us aware of how they integrate or overlap or convert?     (1Q58)

[10:17] Ram D. Sriram: There is a newly formed T ISO/IEC JTC 1 Working Group 10 for IoT. The first kickoff meeting was held on January 27-29, 2015 in Berlin, Germany and one my group members, Eric Simmons, attended this. I am sure he will be glad to brief you all on this.     (1Q59)

[10:17] Ravi Sharma: Q6 We all know that we miss out the relational or predicates power in heirarchies?     (1Q60)

[10:18] Michael Grüninger: Q6: Again, what bridges do we need to build between the IoT working groups and the Applied Ontology community     (1Q61)

[10:18] Steve Ray: Many standards groups develop information models, so one question is whether you would consider a UML model a form of ontology?     (1Q62)

[10:18] Matthew West: Yes. Data models can be a type of ontology.     (1Q63)

[10:19] Steve Ray: @Matthew: I agree, in which case there are lots of ontologies being developed by standards groups.     (1Q64)

[10:19] Sudarsan Rachuri: You may be familiar with http://www.industrialinternetconsortium.org/     (1Q65)

[10:19] Tara Athan: There is new work on giving UML (particularly fUML) formal semantics. This is happening in the OntoIOp committee (OMG).     (1Q66)

[10:19] Adrian Paschke: Standards for Complex Event Processing and Reaction Rules http://link.springer.com/chapter/10.1007%2F978-3-642-24908-2_17     (1Q67)

[10:20] Matthew West: @Steve: Yes there are plenty.     (1Q68)

[10:21] John Morris: When I asked I have shared that ontologies are "scientific conceptual models" -- or depending on how correct one wants to be, one could say "scientific data models". Better software can be expected on the basis of better foundations. Building software not on an ontology is like building a bridge without calculus. It might not fall down, but it's better with science.     (1Q69)

[10:21] Ravi Sharma: @Amanda - request your comment on - data models allow one type of relationships such as PK FK Constraint etc, while ontologies are richer than Data models because they can address multiple types of relationship (attributes and types)?     (1Q70)

[10:22] Liana Kiff: http://linkedbuildingdata.net.previewdns.com/resources/IFC4_ADD1.ttl has been published to formalize the representation of IFC/BIM. But they have left out the annotation from their very rich HTML documentation.     (1Q71)

[10:26] Ravi Sharma: @Amanda -you also mean ontology that relates metadata about data-sets, obviously downstream to sensors and IOT?     (1Q72)

[10:27] Adrian Paschke: ... standardizing the APIs for KBs deployed in distributed IoT environments     (1Q73)

[10:29] Ravi Sharma: @ Tara - OMG Ontology definition metamodel describes how UML is to be profiles and related to RDF OWL.     (1Q74)

[10:32] Mark Underwood: Q7. One of the concerns expressed about standards like AllJoyn, originally started by Qualcomm's Connected Experiences group, is that it's not vendor-neutral. To what extent can an ontology standard mitigate or simply inherit these problems?     (1Q75)

[10:32] Ravi Sharma: @Tara - you are describing post ODM activity I think?     (1Q76)

[10:34] Ravi Sharma: @Mark - Q7 we deal with public ontologies but I guess there could be private vendor implementations that are not open source?     (1Q77)

[10:34] Michael Grüninger: Q7: Ideally, ontologies should be used to mediate and understand differences, rather than to impose semantics     (1Q78)

[10:34] Mark Underwood: Q8. If one or more standards set forth alerting mechanisms, such as the alerting in OGC SensorML 2.0 and the Oasis Common Alerting Protocol used in NIEM and elsewhere, how can ontologies address the problem of alert fatigue? Are there cross-cutting principles from KR that can be applied to guide user interfaces, Big Data variety, information aggregation / concentrators and outputs from predictive analytics?     (1Q79)

[10:35] Gary Berg-Cross: @Tara the last date on the Semantics Of A Foundational Subset For Executable UML Models (FUML), V1.1 seems to be in 2013. Is there any activity with this we should know about?     (1Q80)

[10:36] Michael Grüninger: Q7: Do you mean that the semantics of some concepts within a standard is hidden from users?     (1Q81)

[10:37] Amanda Vizedom: My answer to Q6 is Yes. While the standards groups put a lot of work into producing those taxonomies, etc., those products don't capture well enough the understanding the group may arrive at. Ontologies can capture standards with less ambiguity (thanks to the grounding of logic/set theory/ formal semantics), and in a form ready for machine-interpretation and action. I offer this answer based on experience enhancing the discoverability, and accurate identifiability, of research data sets for NASA; part of this process is, in fact, ontologizing standards to which some data sets are known to conform (I gave the example of the CFMetadata Standards (Climate and Forecast). Without the support of machine reasoning across the ontologized standards and the ontologized data products, it's very hard, and less accurate, for a researcher to find data they can use, even if they know what standard it should conform to. I see a similar dynamic for IoT. If there is a sudden application for, e.g., background radiation measurements that are already being gathered by devices in people's homes and web-accessible, a researcher or emergency responder with use for that data will be greatly helped by such an abstraction layer.     (1Q82)

[10:39] Liana Kiff: Agree with that last comment about alerts. The value of the alert is specific to the person being asked to take action, how, and when.     (1Q83)

[10:39] Tara Athan: OMG has published in the past an attempt at a formal semantics for fUML using Common Logic <http://www.omg.org/spec/FUML/> There are known flaws in what has been published so far, and there is an ongoing effort to fix that in the OntoIOp group <http://ontolog.cim3.net/cgi-bin/wiki.pl?OntoIOp> The problem with put a formal semantics onto a UML model after the fact is that the semantics may not capture what the model developers intended. It is especially important to distinguish between UML models for information objects, which typically correspond to a closed world assumption, and the transference of such a model to the real-world subject of the information object. Consider the case of a model of real-world events versus a data model for an information object that is an event observation. "The finger pointing at the moon is not the moon."     (1Q84)

[10:40] John Morris: In answer to Q on alarm fatigue, I would say there's a big role for ontologies. Modeling systems states can be complex; for example we have a "normal" state where alarms are published to consumers -- but then a degraded or emergency state, which creates "alarm flood", which will overload (as current speaker is saying) humans. I agree with current speaker that programmers or designers can't deal with it. And yes -- "requirements" -- but ontology in and of itself provides the conceptual tool kit for modeling and programming a solution.     (1Q85)

[10:40] Amanda Vizedom: Q8: I agree that ontology *on its own* won't solve this problem. But it can be very helpful, for example by providing a granular, declarative, flexible layer in which dependencies and conditions can be specified, changed, and made sensitive to other information.     (1Q86)

[10:41] Gary Berg-Cross: There are real ontologists involved in semantic sensor network (as you'd expect) and follow ons,but I don't have a sense of which of these other standards efforts being dicussed in passing have ontologists involved. Mark are the ones you involved in have such people and if so who?     (1Q87)

[10:41] John Morris: Amanda Vizedom said it more succinctly. : )     (1Q88)

[10:41] Ravi Sharma: Q8 - the alerts can be generated and validated but in terms of making them reach the desired or affected audience, NIEM like standards would be good facilitators, of course secure and validated alerts only to be shared across communities.     (1Q89)

[10:41] John Morris: The failure to manage requirements (which is partly a failure of project management and investment) is a problem separate from ontology.     (1Q90)

[10:42] Mark Underwood: Q9. A topic touched upon (Hodges, "semantic workflow") but not fully addressed is standardization of workflow and orchestration. In a Big Data, M2M, IoT, sensor-rich world powered by DevOps, what role should an IoT ontology play in marshalling, monitoring, managing IoT resources -- perhaps including (as Ram Sriram suggests) roles performed by human agents? Should ontologies be designed for IoT subsystems following the ontology-oriented design pattern for BPEL suggested by Nitzsche et al (2007) and Aslam et al (2006)?     (1Q91)

[10:42] Ravi Sharma: reverse 911 is a case where a spill notifies the affected community.     (1Q92)

[10:42] Amanda Vizedom: @JohnMorris - Thanks, that's not among the things I hear most frequently. ;-)     (1Q93)

[10:43] Tara Athan: As to putting OntoIOp onto a list of contacts - you might consider two sublists, one for standards directly to IoT and another for standards about supporting technologies. OntoIOp would fit in the second sublist.     (1Q94)

[10:44] Michael Grüninger: Q9: Where are the full references to Nitzsche et al (2007) and Aslam et al (2006)?     (1Q95)

[10:44] Ravi Sharma: Ontology and BPEL can be related for executable ontologies but the mapping is not vewry clear? for Mark     (1Q96)

[10:44] Ram D. Sriram: We definitely need an executive level ontology to deal with the "semantic workflow."     (1Q97)

[10:45] Ravi Sharma: @ram - yes     (1Q98)

[10:45] Bobbin Teegarden: @ram -- yes!     (1Q99)

[10:46] Bobbin Teegarden: @ram, maybe not just an executive level ontology, but actual executable resources integrated INTO ontologies.     (1Q100)

[10:46] John Morris: It's worth noting what apparently are the limitations of BPEL as a graph -- it isn't. If you want freedom to do a process in any arbitrary way, BPEL may not be that model.     (1Q101)

[10:47] Matthew West: Sorry I have to go.     (1Q102)

[10:47] Mark Underwood: ciao Matthew     (1Q103)

[10:48] Adrian Paschke: @John: yes, BPEL is basically orchestration. In IoT it is often choreography style coordination.     (1Q104)

[10:48] Bobbin Teegarden: RE: executable ontologies, what is SPARQL added an E (for execute) the the CRUD is currently does -- CRUDE? -- could you then 'execute' an ontology?     (1Q105)

[10:49] Mark Underwood: @Adrian - maybe BPEL isn't the best framework; that's an issue in Big Data orchestration certainly     (1Q106)

[10:49] Tara Athan: Another group working on supporting technology is API4KB (another OMG RFP) <http://www.omgwiki.org/API4KB/doku.php> We are working on incorporating IoT usecases into this effort (thanks to this summit). We are developing an (OWL) ontology as our metamodel = abstract syntax.     (1Q107)

[10:52] Ravi Sharma: Bobbin and Tara - does OWL allow executable concepts, UML certinly relates to BPMN and thereby worflow but executable?     (1Q108)

[10:52] Ram D. Sriram: Here is what I mean by executive level ontologies. For example, take the traffic situation in any of the major cities. Someone needs to take the sensor data from smart phones, along with the social network information, to make decisions, such as rerouting. One can develop a primitive set of ontologies for this purpose (and this can be used across several domains)     (1Q109)

[10:53] Adrian Paschke: communication in IoT often uses indirect communication. event producers (publishers) and event consumers are decoupled in space and time and communicate via intermediaries such as publish subscribe middlewares. The subscription languages and the event routing mechanisms can make use of ontologies for intelligent routing and more expressive filter operators     (1Q110)

[10:54] Mark Underwood: Q11. What should the role of an ontology (and an ontologist) be when a new generation of sensor devices is introduced -- resulting in not only a mix of sensor streams, but a mix of decision processes, data models and predictability?     (1Q111)

[10:54] Peter P. Yim: MarkUnderwoood: == Q11. What should the role of an ontology (and an ontologist) be when a new generation of sensor devices is introduced -- resulting in not only a mix of sensor streams, but a mix of decision processes, data models and predictability?     (1Q112)

[10:55] Amanda Vizedom: It seems to me that the abstraction layer of ontology should make this easier.     (1Q113)

[10:55] Michael Grüninger: Need to leave now; the recording will continue ...     (1Q114)

[10:56] John Morris: And a budgeting challenge for hospital administrators . . . someone has to fund this episodic cost . . . unless the business model is subscription based, in which case it might work.     (1Q115)

[10:56] Ravi Sharma: Q11 is similar to ERP in wider context, ontologies can be version or change controlled so that we do not introduce ontology for each attribute such as Change anagement, compatibility of Internet explorer is an example of coexistence.     (1Q116)

[10:57] Conrad Beaulieu: I am very interested in evolutionary ontology development with forward and backward compatbility. It seems natural to think that an ontology approach would facilitate this.     (1Q117)

[10:58] Mark Underwood: PDM is involved in this     (1Q119)

[10:58] Bobbin Teegarden: Doesn't change management in an ontology environment drop to the resource, and associated context, level?     (1Q120)

[10:59] Tara Athan: @Ravi and @Bobbin - I have never been completely clear on the terminology regarding "executable ontologies". How is it different from logic programming?     (1Q121)

[11:00] Mark Underwood: Q12. Can a "hard-coded" system (one in which knowledge representation elements are embedded by a developer oblivious to ontologies), be "Ontology Light"-enabled by connecting to various levels or connectors specified by an IoT standard?     (1Q122)

[11:00] Ravi Sharma: @Tara -Static versus dynamic actions based on data or decision rule.     (1Q123)

[11:02] Conrad Beaulieu: Data-driven programming adjusted for ontology interpretation and action. Encoding data schema, process / workflow and execution signatures into an ontology format.     (1Q124)

[11:02] Liana Kiff: Hard-coded systems are dependent upon the stability or the evolutionary capacity of the ontologies that they are "referencing."     (1Q125)

[11:03] Tara Athan: @Ravi - Would an ECA Rule system be consider an executable ontology? <http://en.wikipedia.org/wiki/Event_condition_action>     (1Q126)

[11:04] Ravi Sharma: once hard coded data exist the process of reverse enginnering schema abstraction etc might help undo the damage and then integrate with ontology to remove hard coding disadvantages.     (1Q127)

[11:04] Mark Underwood: Use csae - Wordpress ecosystem for "events"     (1Q128)

[11:05] Mark Underwood: Q13. How is an ontology for decentralized (edge-aggregated or preprocessed streams) different from a centralized one? Should the ontology itself be distributed to the edge? How does this affect software engineering for distributed nodes?     (1Q129)

[11:06] Christopher Spottiswoode: Mark, isn't edge or decentralized a relative matter?     (1Q130)

[11:06] Ravi Sharma: @Tara - studying the link content in the meantime does the event condition consider all "things" and relations to be called ontology?     (1Q131)

[11:07] John Morris: The question of ontology or "semantic distribution" is certainly a technical/performance issue (around bandwidth, CPU power, storage etc.), but also becomes an issue of legal responsibility for data security and business responsibility for service level agreement, at edge, at local area and at enterprise or cloud level.     (1Q132)

[11:08] Christopher Spottiswoode: Yes, ideally they should be relativized.     (1Q133)

[11:08] Mark Underwood: Q15. The ontoCAT integration with R (Kurbatova et al., 2011) for bioinformatics and the rOntorion R package suggest possible collaborations for IoT analytics, perhaps machine learning through R. Microsoft hosted a 2014 conference on machine learning in which a Microsoft blog post argued that "some of the most exciting work being done to reap value from the Internet of Things (IoT) involves taking data insights to the next level using machine learning (ML)." What role should ontologies play in this arena?     (1Q134)

[11:09] Mark Underwood: Adrian - you can enrich a data stream as a producer if you "know" / annotate the consumer's "needs"     (1Q135)

[11:09] Conrad Beaulieu: As much of the ontology as possible should be pushed to the edge - even if it is used for higher level devices to reference about the device. Subseting the ontology for device specific definitions which can be stored in cheap memory - even in small devices. Binary annotation can be mapped to the ontology definition.     (1Q136)

[11:09] Ravi Sharma: @Tara - YES Link is good example - It can be executable ontology related to our earlier determination of Ontology and BPEL or excutable, this is very relevant example.     (1Q137)

[11:10] Mark Underwood: Q - What is our Ontology at the Edge history?     (1Q138)

[11:10] Mark Underwood: Q16. Would the presence of a readily usable ontology for a class of IoT devices facilitate white hat / black hat IoT attacks, penetration testing?     (1Q139)

[11:10] Ravi Sharma: you mean Netcentric DOD edge?     (1Q140)

[11:11] Amanda Vizedom: @MarkUnderwood -- I could have missed it, but I don't recall much if any our discussions being framed specifically in terms of the "the edge" (DoD sense or otherwise).     (1Q142)

[11:12] Christopher Spottiswoode: Q16: Certainly, Access Control ontologies should be available for reuse.     (1Q143)

[11:12] Tara Athan: @Ravi - so taking "ontology" in a general sense (not just OWL), ECA Rules would be an executable ontology that focuses on rules, while other executable ontology languages might have a different emphasis. OWL itself can be used to model and describe events, conditions and actions, but not to execute them.     (1Q144)

[11:13] Mark Underwood: Q17. Are there principles from the 2008 Ontology Summit Metadata for Ontologies discussion that should be reinvigorated for IoT settings?     (1Q145)

[11:14] Ravi Sharma: @Tara - but your link prescribes execution conditions, can OWL incroporate this as something similar to UML Profile with action attribute?     (1Q146)

[11:15] Ravi Sharma: @mark - Yes     (1Q147)

[11:15] Ravi Sharma: on Q17, yes     (1Q148)

[11:16] Christopher Spottiswoode: The role of standards? Subsidiary to component architecture and integrated reuse tools.     (1Q149)

[11:16] Tara Athan: @Ravi - OWL has a fixed model-theoretic semantics. It is inherently passive.     (1Q150)

[11:19] Christopher Spottiswoode: Re my answer above re standards, component architecture must encompass metadata.     (1Q151)

[11:20] Adrian Paschke: @Ravi: in general there is a difference between active execution of events and actions and reasoning about the effects of events/actions.     (1Q152)

[11:20] Christopher Spottiswoode: Yes Mark, it's very key!     (1Q153)

[11:21] Adrian Paschke: OWL is monotonic     (1Q154)

[11:21] Ravi Sharma: @Tara - then using OWL to UML2 with BPMN and I am hoping BPEL route can provide an action option in ontology context?     (1Q155)

[11:22] Ravi Sharma: We want to thank Mark for great interaction opportunity.     (1Q156)

[11:22] John Morris: Nice moderation in a dynamic environment!     (1Q157)

[11:22] Peter P. Yim: good session, Mark ... thank you for organizing this!     (1Q158)

[11:22] Liana Kiff: Thanks, Mark!     (1Q159)

[11:22] Peter P. Yim: -- session adjourned --     (1Q160)

[11:22] Christopher Spottiswoode: Mark, very well handled session - thanks!!     (1Q161)

[11:24] Mark Underwood: Link to Ontocat for "R" package for ontology traversal and search http://bioinformatics.oxfordjournals.org/content/early/2011/06/22/bioinformatics.btr375.full.pdf     (1Q162)

[11:25] Mark Underwood: Link to Microsoft's blog post on IoT and analytics: http://blogs.microsoft.com/iot/2014/12/09/machine-learning-adding-impact-to-iot/     (1Q163)

[11:26] Amanda Vizedom: Re: whether there is a session next week: Summit Sessions are among the events listed on the Ontolog Events calendar https://www.google.com/calendar/render?cid=Y2RqMWxia241ZzFyN2dpMWwxZmZkMm1rZWNAZ3JvdXAuY2FsZW5kYXIuZ29vZ2xlLmNvbQ     (1Q165)

[11:26] Amanda Vizedom: ... and answer is yes, There is a Track B Session. :-)     (1Q166)

[11:28] Mark Underwood: Nitzsche - BPEL for Semantic Web Services http://link.springer.com/chapter/10.1007%2F978-3-540-76888-3_37#page-1     (1Q167)

[11:29] Mark Underwood: Aslam et al on Expressing Business Process Model as OWL-S Ontologies http://ro.uow.edu.au/infopapers/439/     (1Q168)

[11:30] Mark Underwood: Compton et al - Sensor Data Provenance SSNO and PROV-O http://knoesis.org/ssn2014/paper_9.pdf     (1Q169)

[11:33] Mark Underwood: Thanks, all     (1Q170)