ppy/chat-transcript_unedited_20120315b.txt ----------- Chat transcript from room: summit_20120315 2012-03-15 GMT-08:00 ----------- [09:18] PeterYim: Welcome to the = OntologySummit2012: Session-10, Thursday 2012-03-15 = Summit Theme: OntologySummit2012: "Ontology for Big Systems" Track (3) Title: Challenge: Ontology and Big Data Session Topic: Big Data Developing Challenges Session Chairs: Ms. MaryBrady (NIST) and Mr. ErnieLucier (NCO/NITRD) Panelists: * Professor TimFinin & Professor AnupamJoshi (UMBC) - "Making the Semantic Web Easier to Use" * Dr. KyoungsookKim (NICT, JP) - "Use cases of cyber-physical data cloud computing" * Dr. MikeFolk (HDF Group) - "The HDF5 technology suite" * Dr. MarioPaolucci (LABSS/ISTC/CNR, Rome, Italy) - "FuturICT: Global Participatory Computing for Our Complex World" * Dr. UrsulaKattner (NIST) - "Data Needs for the Materials Genome Initiative (MGI) at NIST" * Dr. EdinMuharemagic (HPCC Systems; LexisNexis) - "HPCC Systems Machine Learning" Session page: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2012_03_15 Mute control: *7 to un-mute ... *6 to mute Can't find Skype Dial pad? ... it's under the "Call" dropdown menu as "Show Dial pad" . == Proceedings: == . [08:35] anonymous morphed into Tim Finin [09:21] anonymous morphed into Hasan Sayani [09:23] anonymous morphed into MatthewHettinger [09:25] BobSchloss: Peter, Leo et al: I am thinking about the April 12-13 F2F at NIST. I may not be able to be in Maryland by first thing in the morning on April 12th. Would you aim to put some details about start time, and agenda, on the page for the Symposium sometime in the next week. Thank you [09:26] anonymous morphed into MikeFolk [09:27] BobSchloss: I do see that the page says that if you arrive at NIST at 8am, there will be time to get through security before things start. I just want to check that I could still arrive later and get through Security.... If there is one person I should talk to about such logistics, just give me their name, e-mail address, phone number. Thank you. [09:29] anonymous1 morphed into Ed lowry [09:30] anonymous morphed into Andrea Westerinen [09:32] anonymous morphed into ChristopherSpottiswoode [09:32] Kyoungsook Kim: can you here me [09:32] anonymous1 morphed into Mario Paolucci [09:33] anonymous morphed into David Orloff [09:35] PeterYim: == Chairs doing intro ... [09:38] anonymous morphed into CoryCasanave [09:38] PeterYim: == TimFinin presenting ... [09:42] MaryBrady: Bob, you can most certainly arrive later at NIST and get through security. Arriving at 8:00 will just allow you to make it through prior to the start of the meeting. [09:46] DougFoxvog: I've been unable to connect to Shared-screen support at http://vnc2.cim3.net:5800/ . Is it up and running? [09:48] MichaelGruninger: @BobSchloss: We will be putting an initial agenda up on the Symposium page sometime tomorrow [09:48] PeterYim: @Doug ... it should be - does anyone else have problems (especially if you have been able to do that before) [09:50] anonymous morphed into Rosario Uceda-Sosa [09:51] Mario Paolucci: I'm seeing the VNC but it took quite a while to establish connection [09:52] anonymous morphed into ElizabethFlorescu [09:52] PeterYim: @Mario - don't worry about it (if you cannot get a "good" connection to the vnc server) just use your own slides (on your desktop) but please remember to call out slide advances and call out the slide numbers each time [09:54] PeterYim: @MaryBrady - we have at least 10 people on voice (conference bridge) who are not in the chat-room ... could you remind people to join us in the chat-workspace, the next time you have a chance, please [09:55] MaryBrady: @PeterYim, will do [09:55] PeterYim: Tx [09:58] DougFoxvog: I was unable to connect with Firefox. But i managed to dredge up an Internet Explorer, and there was no problem. If anyone else is having a problem connecting with Firefox, i would suggest they try a different browser. [09:59] PeterYim: @BobSchloss - ref. person you can call - the official contact I have from the NIST registration page is at: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012/WorkshopRegistration#nid36LP ... so feel free to call me on my mobile ( 650 578.9998 ) too [10:02] anonymous morphed into JosephTTennis [10:03] anonymous1 morphed into JodyDesRoches [10:03] anonymous2 morphed into Buck Nimz [10:03] anonymous morphed into Ursula Kattner [10:03] DeborahMacPherson: Hi Everyone! Interesting discussion this all has been. I have several entire threads set aside to read again fully [10:03] anonymous1 morphed into ChristopherSpottiswoode [10:04] Kyoungsook Kim: hi [10:04] MaryBrady: Dr. Kim, we are ready, *7 to unmute [10:04] SteveRay: @Tim Finin: Very cool. Is there a way to point your GOR tool to other SPARQL endpoints? [10:04] PeterYim: == KyoungsookKim presenting ... [10:08] PeterYim: @MaryBrady - if there is some way you can turn down your mike volume a little, that would be nice (as the level is a little high now, and it some time clips when you speak) ... if you cannot turn down the volume, please try not to get too close to the mic [10:08] MaryBrady: @PeterYim, I've turned it down...let me know if it is ok [10:10] PeterYim: @Mary - great, thank you! ... will do (after you speak next time) [10:10] LeoObrst: @Tim: concerning Varish Mulwad's research (inferring semantics of tables), was Formal Concept Analysis considered? [10:13] Tim Finin1: @leo --- No, not to my knowledge. Good idea. We'll look into it and think about its value for this problem. [10:13] ErnieLucier: HTTP copies as hFp for all reference URLs in Tims presentation. Replacing hPf with http works. [10:16] JosephTTennis: This has been GREAT! Sorry I was late. And sadly, I'll have to leave early :-( Looking forward to participating more in the near future! [10:16] PeterYim: @JosephTTennis - Thank you for the participation, Joe [10:17] JosephTTennis: ciao! [10:20] NicolaGuarino: Folks, unfortunately I have a car emergency, I have to leave. Perhaps I'll manage to connect again in 40 minutes or so, not sure. Sorry missing Mario's presentation. [10:21] PeterYim: @Nicola - thank you for the heads up ... bye! [10:21] ErnieLucier: Is permission to use social networks required? [10:23] PeterYim: == MikeFolk presenting ... [10:28] Harold Boley: @Dr. Kim, in Real-world Awareness Computing, could the Observation/Perception/(Communication)/Action sequence be formalized using Event/Condition/Action rules? [10:32] Kyoungsook Kim: currently, I try to use a rule-based language. [10:32] Kyoungsook Kim: like datalog [10:34] Harold Boley: Maybe the premises of datalog rules need to be partitioned into Event and Condition parts? [10:35] Harold Boley: Events are 'sensed' as external observations. Conditions 'test' the internal knowledge base. [10:39] BobSchloss: Just an observation -- many people are working on highly scalable triplestores, some with interesting partioning and distribution and federation functions. We might sometime convene a panel with all of these people showing what they did. It is not just graph stores that are advancing -- the entire NoSQL movement is starting to develop various interesting strategies, in which some of the classic ACID properties are slightly relaxed. [10:41] AliHashemi: The pdf version of this presentation is not rendering appropriately for me. Is it just me? [10:41] Kyoungsook Kim: me too [10:42] AmandaVizedom: @MikeFolk - HDF5 is new to me, so I find I have some "what *is* HDF5?" questions below the level of your talk. I see that at http://www.hdfgroup.org/HDF5/, there are links for "What is HDF5?" "Questions (FAQ)," and "HDF5 Tutorial" links. Would you recommend these as the best source for an overview of the fundamentals of HDF5? [10:45] AliHashemi: @Kyoungsook - I downloaded the file and opened it in Reader - it works ok there, I think my browser's pdf reader doesn't show it correctly. [10:45] PeterYim: == MarioPaolucci presenting ... [10:48] Kyoungsook Kim: thank you. [10:55] David Orloff: Sorry need to run to a meeting - if anyone is looking for data to work with that is already tagged with ontologies (9 in use) look at www.cellimagelibrary.org and contact me David Orloff at dorloff@ascb.org - thanks for letting me sit in - I will be back for future calls. [10:55] PeterYim: @MarioPaolucci - how does the project mitigate between its desire to be "open" versus it's dependencies (ref. your slides) on commercial products (like skype, facebook, etc. ... which are usually non-open) [10:56] Harold Boley: @MarioPaolucci, what would be the initial steps for moving from a Strongly Coupled System to a Weakly Coupled System? [10:57] PeterYim: == UrsulaKattner presenting ... [10:59] Mario Paolucci: @PeterYim: There are several strategies possible. We plan to build alternative data sources and provide access to them. We will not depend on commercial platforms (the names in the slide were there more as an examples of changes brought about by technology), but we hope to create our own data sharing platform, where user themselves authorize access to their data. In a sense, we hope to convince people to reclaim access to their data. [11:01] Mario Paolucci: @Harold: I can only provide stylized examples; it depends from the specific problem. But of course you can hardly intervene on the self-organization, so what can be done is changing the terrain where things happen. The examples that come to my mind is the roundabout instead of the intersection; or in a sand pile model, breaking up the table so that cascades remain limited. [11:04] DeborahMacPherson: Interested in discussing the differences between calculated and measured values referred to by the speaker just now [11:05] FrankOlken: @PeterYim Will all of the Ontology Summit sessions be webcast? I am thinking of attending remotely ... [11:07] Harold Boley: @Mario, You could 'overlay' the roundabout -- with 4 quarter-circle 'bypasses' -- over the intersection, so at least to help those not in the center of the congestion. [11:08] Mario Paolucci: @Harold:Exactly. Also, adding new dimensions help - either by digging a tunnel, or - better - providing car with vertical mobility. Think if this as a metaphor - new dimensions are easier to create in virtual worlds, of course [11:10] Mario Paolucci: ops. I forgot to put the links. [11:10] Mario Paolucci: list of supporters: http://www.futurict.eu/the-project/whos-involved [11:10] DeborahMacPherson: Need to sign off - thanks to all the speakers - [11:11] Mario Paolucci: to join: http://www.futurict.eu/the-project [11:11] PeterYim: == EdinMuharemagic presenting ... [11:11] DougFoxvog: I think we're on slide 3 [11:11] MaryBrady: Please, if you have questions, post them here...we'll be sure to engage the speakers in answering the questions over e-mail [11:12] PeterYim: @Mary - please ask the speaker to call out the slide advance AND slide number [11:12] Ursula Kattner: The link for the Materials Genome Initiative is http://www.whitehouse.gov/blog/2011/06/24/materials-genome-initiative-renaissance-american-manufacturing [11:16] PeterYim: @FrankOlken - are you referring to the OntologySummit2012_Symposium (at NIST on 4/12 & 13)? [11:16] MaryBrady: In particular a number of technologies and use cases for BIG DATA have been presented this afternoon. Any thoughts on potential uses for Ontology? [11:17] PeterYim: @Frank - ... assuming that, the answer is "yes" - remote participation will be supported for all the sessions [11:17] FrankOlken: @PeterYim Yes, I am referring to the Ontology Summit on April 12-13 at NIST. I am already committed to a trip the previous week (Data Engineering conf) and am [11:18] FrankOlken: @PeterYim .. am reluctant to commute to DC twice in 2 weeks. [11:19] FrankOlken: @PeterYim I think that your infrastructure for supporting the teleconferences has worked quite well. I use skype to listen. [11:20] PeterYim: @Mario - thank you ... but then, the challenge comes in the form of how (and if at all) one can build out a user base of hundreds of millions of people (like the success some of these commercial social network platforms have achieved) [11:21] anonymous morphed into NicolaGuarino [11:21] Mario Paolucci: @Mary: We have ontology components in all parts of the FuturICT architecture, of course. Nicola Guarino knows more about them. But we have a critical need of ontologies that allows the different components to communicate - think of aligning models and simulation results along disciplines (sociology, complex science) and along levels of detail (individual agents, organizations, groups, etc.) There should be a world of modeling component in which ontology is very important. [11:21] Mario Paolucci: @Nicola: bentornato! [11:22] SteveRay: @Frank: We may have a problem using Skype when we are at NIST, because they ban Skype there for security reasons. [11:23] MaryBrady: Any thoughts on the integration of ontology components with output from machine learning techniques? [11:24] PeterYim: @Steve, @Frank - we will be hosting calls in a way (and with the same tools) similar to all Ontolog virtual session ... except that we may not support shared-screen (vnc), but then, that has never been a show stopper for us [11:24] FrankOlken: @MaryBrady I recall that some folks have suggest using ontologies to suggest concepts to be learned. [11:26] MaryBrady: @FrankOlken Yes, here at NIST we have used combination techniques between ontologies and machine learning. Simple queries can sometimes take days to complete. [11:28] Mario Paolucci: @PeterYim: We think users would be happier to share data with a privacy preserving, non profit project, but it's a risky bet, I agree. [11:34] Ursula Kattner: @DeborahMacPherson: Measured values have a confidence resulting from the error of the measurement, calculated values have no such error. However, a confidence for these data is needed to properly judge them in the context of data that describe a material. [11:35] LeoObrst: Must go now. Very interesting session. Thanks to all! [11:35] MatthewHettinger: @Edin, What open source products are you using? [11:35] DougFoxvog: @Mario You discuss using "Crowd sourcing" and "citizen science" for a platform for economic and political participation. People on different sides of various issues would have competing "science". How would you deal with this? [11:38] NicolaGuarino: @Doug: here is exactly one of the roles of ontologies in this project: exposing disagreements about different opinions... [11:39] PeterYim: @KyoungsookKim - how effectively did the systems cited in your use cases turn out (in real life) ... were there metrics available? [11:40] NicolaGuarino: @Doug: the point is *understanding* the different models, not necessarily forcing them to align one each other [11:40] AmandaVizedom: Dr. Kim, someone responded to my G+ posting about your presentation by mentioning evacuation response research such as some at Univ. of Minnesota ( http://gradworks.umi.com/32/05/3205248.html, http://www.spatial.cs.umn.edu/paper_ps/evac_SSTD05.pdf). My response is that this sort of research, in itself valuable, would relate to your use case as a contribution to *one* of the areas of computation involved in the response. It seems to me that what makes your use case such a Grand Challenge type case is that it brings together a variety of such areas, including route-planning and information fusion across very heterogenous sensor and information types and disaster surveillance over networks. Do you agree? [11:42] AmandaVizedom: @KyoonsookKim - I should add that I think it's really a very good Grand Challenge, for a few reasons, including that it is so well grounded in a real need *and* real, existing data environments, and success has such clear benefits. [11:43] ErnieLucier: @Dr. Kim, Is permission to use social networks required or a problem? [11:43] DougFoxvog: @Nicola So long as the different theories/models are kept separate, i strongly agree. The problem i saw was with an "open" system which would allow people to modify theories that they didn't create. [11:44] SteveRay: Must run. Thanks for a stimulating session. [11:44] Mario Paolucci: Thank you everybody for listening and for the questions. I have to leave now, bye! [11:45] NicolaGuarino: @Doug: You are right. Definitely people shouldn't be allowed to modify things at their ease... especially if the underlying assumptions are not shared... [11:46] Kyoungsook Kim: using social network, we don't try to use personal information [11:48] Kyoungsook Kim: itself. We aggregate a group of messages and extract trend information or changing information. [11:48] FrankOlken: Nearly every machine learning algorithm is available under R. [11:49] FrankOlken: There are versions of R that run on clouds with Hadoop. [11:49] PeterYim: wonderful session ... great presentations! [11:50] FrankOlken: There is recent work at IBM and Univ. of Wisconsin on parallel implementation of stochastic gradient descent for machine learning. [11:50] PeterYim: -- session ended: 11:49am PDT -- -----------