ppy/OntologySummit2013_s08_chat-transcript_edited_20130307b.txt ------ Chat transcript from room: summit_20130307 2013-03-07 GMT-08:00 [PST] ------ [09:26] PeterYim: Welcome to the = OntologySummit2013: Virtual Panel Session-08 - Thu 2013-03-07 = Summit Theme: Ontology Evaluation Across the Ontology Lifecycle * Summit Track Title: Track-A: Intrinsic Aspects of Ontology Evaluation Session Topic: Intrinsic Aspects of Ontology Evaluation - II * Session Co-chairs: Dr. LeoObrst & Dr. StevenRay Panelists / Briefings: * Professor PatrickLambrix (Linkping University, Sweden) - "Debugging is-a structure in ontologies" * Ms. MariaCopeland (University of Manchester) - "Ontology Evolution and Regression Testing" * Dr. MelissaHaendel (Oregon Health & Science University) - "A biologists' perspective on ontology utility" * Mr. EdBarkmeyer (NIST) - "Core components for an ontology: Modeling Codes and Code Lists" Logistics: * Refer to details on session page at: http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2013_03_07 * (if you haven't already done so) please click on "settings" (top center) and morph from "anonymous" to your RealName (in WikiWord format) * Mute control: *7 to un-mute ... *6 to mute * Can't find Skype Dial pad? ** for Windows Skype users: it's under the "Call" dropdown menu as "Show Dial pad" ** for Linux Skype users: please note that the dial-pad is only available on v4.1 (or later or 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. Attendees: AliHashemi, AmandaVizedom, AnatolyLevenchuk, AstridDuqueRamos, BijanParsia, BobSmith, BobbinTeegarden, DavidWhitten, DennisWisnosky, DmitryBorisoglebsky, DougFoxvog, EdBarkmeyer, FabianNeuhaus, FranLightsom, HansPolzer, JoanneLuciano, JoelBender, JohnBilmanis, KenBaclawski, KevinSimkins, LeoObrst, LudgerJansen, MariaCopeland, MeganKatsumi, MelissaHaendel, MichaelGruninger, MikeDean, PatrickLambrix, PavithraKenjige, PeterYim, RamSriram, SteveRay, TerryLongstreth, ToddSchneider, TorstenHahmann, ValentinaIvanova, VictorAgroskin, vnc2 . == Proceedings: == . [9:09] anonymous morphed into PatrickLambrix [9:23] anonymous morphed into KevinSimkins [9:29] PeterYim: Attn ALL: ... it has come to our attention that our conference bridge provider is running into some problems with the "joinconference" skype connections. In case anyone gets in trouble, please try to call the phone numbers instead (e.g. from your phone, skype-out, google-voice, etc.) [9:30] anonymous morphed into MelissaHaendel [9:30] KevinSimkins: The IEEE Virtual World Standard Working Group (P1828) is focused on the development of common standards for virtual environments. From assets to protocols and general working models in order to facilitate the common interoperability scenario for all virtual environments in the future. Our working group members are dedicated to multiple phases within this process in order to accumulate further details and application methodologies over time. ... see: http://www.metaversestandards.org/index.php?title=Main_Page [9:32] JoelBender: @Peter - attempting to enter the conference ID is failing, the bridge is missing digits - I'm calling from a land line [9:38] AnatolyLevenchuk: wrt skype: always mute microphone when you enter ID to skype (mic take sound from your tone-digits and then can be "strange" processes of echo and digits suppressing or doubling of click and "listened" digits). [9:34] anonymous morphed into BobSmith [9:35] MelissaHaendel: yes, screen share doesn't work for me [9:36] FabianNeuhaus: Melissa, the screen share might not work if you are behind a firewall [9:37] FabianNeuhaus: just download the slides from the session page http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2013_03_07 [9:39] MelissaHaendel: thanks Fabian, I have the slides [9:35] anonymous morphed into TorstenHahmann [9:38] PeterYim: == SteveRay opening the session on behalf of the co-chairs ... ... see: the [0-Chair] slides [9:41] List of members: AmandaVizedom, AnatolyLevenchuk, AstridDuqueRamos, DennisWisnosky, EdBarkmeyer, FabianNeuhaus, FranLightsom, HansPolzer, JoanneLuciano, JoelBender, JohnBilmanis, KevinSimkins, LeoObrst, MariaCopeland, MelissaHaendel, MichaelGruninger, MikeDean, PatrickLambrix, PeterYim, RamSriram, SteveRay, TerryLongstreth, ToddSchneider, TorstenHahmann [9:41] PeterYim: == PatrickLambrix presenting ... see: the [ 1-Lambrix ] slides [9:44] anonymous morphed into DougFoxvog [9:51] MelissaHaendel: Question, so NCI is largely a human representation whilst MA is mouse. How do you know there are not species-specific is_a absences or presence? Examples are correct, though :-). [11:24] PatrickLambrix: @Melissa [9:51]: For our experiments we had the help of a domain expert. (This doesn't mean all the validations are correct. We have also noticed that a domain expert may change his/her mind about the correctness of an is-a relation during the debugging session when more is-a relations were debugged.) [11:28] MelissaHaendel: @patrick. I understand biologists changing their mind! Also depends on the way in which they are viewing the ontology and the debugging changes. Would be interested to view this sometime, will look at your papers. [9:54] SteveRay: @Melissa: My understanding from his talk is that is why the repairs must be validated by a domain expert. [11:24] PatrickLambrix: @Steve [9:54]: Yes, repairs need to be validated by a domain expert. For missing is-a relations a system can compute repairs that guarantee that the missing is-a relations will be logically derivable form the repaired ontology, but this does not necessarily mean that the repair is correct according to the domain. [9:54] AmandaVizedom: @PatrickLambrix - I like your focus on a high-level division into syntactic, semantic, defects; your calling out that detection and debugging of the modeling defects requires domain knowledge, and your subsequent emphasis on very practical ways to detect and debug this third category. In my experience, the syntactic and semantic evaluation and correction, while necessary, are rarely sufficient. There seems, however, to be a widespread impression, or fear, that practical, reliable evaluation of model accuracy is not possible. Thanks for showing this not to be true. [11:24] PatrickLambrix: @Amanda [9:54]: If I remember well, the different kinds of defects were defined in Aditya Kalyanpur's PhD thesis. However, there is not so much work on systems for the modeling defects yet. [9:58] DougFoxvog: For the example, concluding "limb_joint" is-a "joint" because all of its subclasses are-a joint although it is not in any of the ontologies, the conclusion must be verified by a human. It is quite possible that that common is-a is too general a term -- which may be the reason the is-a was not asserted. [9:58] PeterYim: @PatrickLambrix - are there tools already implemented for the process outlined in slide#26 ... if so, are they openly accessible (what is url?) [9:58] MelissaHaendel: It would be great to have tools that present these things easily to the domain expert. Actually, I would not classify a hip joint as a limb joint (sorry, I work on this stuff ;-)). [11:24] PatrickLambrix: @Peter,@Melissa: [9:58]: We have systems and plan to make the first system available soon. There are screenshots in the papers in the reference list of the talk. [10:05] DougFoxvog: @Melissa: That depends upon the ontology's definition of "limb joint". The term could mean a joint within a limb or a joint which is to at least one bone in a limb. Bone joints vs. body region joints are also a contrast; these two types are disjoint. [10:01] AmandaVizedom: @PatrickLambrix - I would note that "is-a" in your examples appears to be a subClass or kind-of relationship (or, if also used for instances, it may be an under-specified narrower concept used for both instancehood and subclass relationships, as found in some taxonomies, less commonly ontologies, and other semantic models). I call attention to this simply because it can be confusing to modelers who follow most contemporary ontology languages, in which "is-a" is used only for the instancehood relationship, and a different relationship (subClassOf, kind-of, #$genls) is used for the sub-class relationships in your examples. [10:06] MelissaHaendel: well, many anatomists define limb as the free limb, as opposed to the limb plus girdle. but that is perhaps neither here nor there. [10:06] MelissaHaendel: @Patrick - I would love to have you experiment with our work on Uberon - see http://uberon.org. [11:25] PatrickLambrix: @Melissa [10:06]: I would be interested at looking at Uberon. [10:09] DougFoxvog: @Amanda: many people use "is-a" to mean subclass of. Cyc uses #$isa to mean instance of. I (growing up with Cyc) also find "is-a" used for subclass-of as grating. Grammatically, it should be "a-is-a". [11:24] PatrickLambrix: @Amanda,@DougFoxvog [10:01]: Yes, is-a isSubClass or kind-of. Our system works currently only on the concept level. For is-a we assume relexivity and transitivity. [10:10] AmandaVizedom: @doug, far fewer than used to. The two relationships are teased out, for example, even in OWL. [10:05] JoanneLuciano: is is-a a kind-of-kind-of is-a relationship? or is it a-kind-of-like relation? [11:24] PatrickLambrix: @Joanne [10:05]: exactly :-) [10:05] JoanneLuciano: :-) nice presentation! [10:10] LeoObrst: @Patrick: do repair actions ever cause additional problems for other parts of the ontologies not yet analyzed? [11:25] PatrickLambrix: @Leo [10:10]: Repairing actions could cause additional problems. Therefore, the system allows a user to switch between the different phases of detecting, validating and repairing, as well as taxonomies and iteratively debug the whole network. The system also checks whether a domain expert would want to remove a previously validated to be correct is-a relation. In this case the user needs to retract the previous choice or the repair is not allowed. Similar when the domain expert wants to add an is-a relation, that was previously validated as wrong. [10:06] PeterYim: == MariaCopeland presenting ... see: the [ 2-Copeland ] slides [10:14] EdBarkmeyer: @maria: Only some tested elements have meaning to the user. Many test elements are intended to validate that the inferences work as expected for known cases. In many cases, the user cannot recognize an erroneous inference, as Patrick mentioned. [10:21] EdBarkmeyer: @maria: Is it clear that these changes are "truth" and "bugs", or just differences in educated opinion prevailing at different times in the maintenance cycle? [10:21] AliHashemi: @Maria - are the graphs on slides 30 and 34(b) correct? It seems to me it should be Effectually added (solid line) Ineffectually Removed (dotted line) Effectually Removed (no line) Effectually Added ... otherwise, what do the lines represent? [10:23] DougFoxvog: @Maria: could these changes be a result of different users having different ideas of what the terms (should) mean? [10:23] MelissaHaendel: We (try to) keep track of such edits in the ontology metadata, it would be fantastic to have a human readable version provided to the ontology editor/domain editor whilst they are editing. [10:23] anonymous morphed into DavidWhitten [10:26] AmandaVizedom: @MariaCopeland - could you explain your effectually / ineffectually distinction a little more? I think I understood you to say that "effectual" presence means presence as a directly asserted axiom, while "ineffectual" presence means absence from the directly asserted axioms, but continued entailment by the ontology. Is that correct? If so, is this, as the label "ineffectual" suggests, considered a fault? I ask because I imagine that presence as entailment only would be desirable in some applications and undesirable in others. Specifically, if little to know reasoning is done in an application and only directly asserted applications are usable, then entailments are indeed "ineffectual". In other cases, where some degree of reasoning is used and is efficient, assertions that are redundant with entailments may be removed for efficiency, if there is not additional, provenance-related reason to make explicit assertions that rely on different sources. [11:24] MariaCopeland: @EdBarkmeyer, @AliHashemi, @DougFoxvog, @Melissa: Thanks for your questions and comments. I will follow up offline. [10:23] PeterYim: == MelissaHaendel presenting ... see: the [ 3-Haendel ] slides [10:30] HansPolzer: (ref. slide#4) The example of fruit fly limbs vice human limbs underscores the need for more explicit representation of context as it relates to ontologies [10:32] DougFoxvog: @Hans: the fruit fly vs. human Tibia presents an example of using NL words as terms for concepts in an ontology. There should be mappings from the NL terms to (multiple) concepts. But annotation of the concept should make its intended meaning crystal clear. [10:34] AmandaVizedom: @Hans - I don't disagree, exactly, but I'd add that it is also a good illustration of why it is important not to confuse expressions and concepts. Arguably, the fruit fly tibia is a subclass of some class of animal body parts that are not skelatal; the human tibia is a subclass of some class of animal body parts that are skelatal. "limb" may be an expression used for both of those classes, but the classes are different. [10:35] DougFoxvog: @Melissa : Having a key separation in an ontology whether something is "detectable" seems to be setting oneself up to have the feature expire when new technology allows something not previously detectable to become detectable. [10:36] DougFoxvog: I had trouble hearing for a while. But the sound has now come back [10:36] HansPolzer: But in some contexts the term "limb" represents the identical concept, for example, when considering locomotion rather than anatomy (and not focusing on how exactly the movement of the limb is activated/energized.) [11:02] MelissaHaendel: @hans. we have disjoint axioms to prevent fruitfly tibia being equated with human tibia. I wonder how much such disjoint axioms are leveraged in the ontology alignment strategies? or if we can be providing them better for such purposes. [11:04] DougFoxvog: +1 Melissa. Disjointness axioms are crucial. Especially if they are enforced on modification of an ontology to prevent the creation of classes that are disjoint with themselves. [10:37] AmandaVizedom: @Melissa -- Don't worry, it's not just biologists! I've lost track of the number of domains I've worked in or with, and I don't think I've yet met a community of practice that doesn't have these features (moving conceptual targets, reuse of labels, and expert disagreements on definitions). I suspect there are fundamentals of human cognition, language, and community knowledge development at work . :-) [10:38] DougFoxvog: One need to define multiple limbs: tree limb, animal limb, subclasses of animal limb by exo/endo-skeleton, etc. [10:38] AmandaVizedom: @Hans - agreed. And that is, arguably a different concept, for which the same expression is used. [10:39] DougFoxvog: @Amanda: Of course. Replace disagreement on definitions by modeling all definitions with different (maybe overlapping) concepts. [10:39] SteveRay: This raises the point that different communities still want to use their own terminologies, and they should, but we can separate those terminologies (expressions or labels) from the concepts. [10:41] EdBarkmeyer: @melissa: As you say, people think ontology terms are words, and they think definitions are circular when those words are reused, but the fact is that the definitions are in natural language, the ontology symbols are "words" in a formal language. The issue is whether the definitions themselves are circular. [10:42] DougFoxvog: OBO still has problems with different people having different ideas of what a term is, even when the name is numeric. The NL description is not specific enough, so for example, the plant people assume it means the plant definition of the term and the animal people assume it is an animal definition. As a result, in the Cell (Line) Ontology, multiple cell types were subclasses of both animal cell and plant cell. [10:41] TerryLongstreth: @Melissa - Last bullet on slide 11 implies that speculative biology is not in scope. [11:04] MelissaHaendel: @terry I think what I was saying is that actually speculative biology is in scope, but we just need to understand how to apply the ontology consistently when we are potentially being speculative. (or maybe I misunderstand) [10:41] HansPolzer: We have an aversion to specifying context because it seems unnecessary in most situations we find ourselves in - yet the internet and networked world exposes us to otherwise alien contexts that are left implicit, thus leading to the arguments about definitions [10:42] HansPolzer: We want an absolute frame of reference, but usually assume our own frame of reference is that absolute frame of reference. [10:44] HansPolzer: We need to learn to tolerate that there are multiple frames of reference and multiple perspectives on those frames of reference and associated scope. [10:44] DougFoxvog: @Hans: You say "we" have an aversion to specifying context. *I* don't. 8)# I find context crucial. [10:45] HansPolzer: It was the editorial "we" :-) [10:45] HansPolzer: Humans are very good at detecting context, except when it fails us - and thus results in great literature [10:46] HansPolzer: and puns [10:43] anonymous morphed into ValentinaIvanova [10:44] SteveRay: Gee, I'm thinking about a transiently transfected DNA expression construct and I'm coming up blank :) [10:45] AmandaVizedom: @Melissa - slide 16: very nice illustration of refinement to include more specific classes, corresponding with important functional distinctions. [10:46] EdBarkmeyer: @melissa: this is a great contribution! thanks. [10:50] PeterYim: @MelissaHaendel: do you have tools to augment the process(es) that you are describing [11:07] MelissaHaendel: @Peter - we have some scripts that help. We have been using Jenkins to help too. [11:12] PeterYim: @MelissaHaendel - thank you, reason I am asking is because we are running "Ontology Clinic" activities (as part of this Summit) by putting together ontology developers and people with ontology evaluation methodologies and tool in the same (virtual) room to get some interesting outcome, and we'll love to have you (and the other panelists today) to join us - see: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2013_Hackathon_Clinics [11:17] MelissaHaendel: @peter, it would be great to see if you could get Chris Mungall to participate regarding his development of Jenkins. see http://wiki.geneontology.org/index.php/Jenkins . Most of our work on the metadata is in the form of scripts and not integrated into tooling (though we have some protege plugins). [11:19] PeterYim: @Melissa - thank you for the pointer ... this is helpful as we are doing a survey on software and tools, as part of the Summit activities, too [10:47] DougFoxvog: Re slide 15. An ontology should not have synonyms. It should have mappings from NL terms to the ontology terms. One NL term can map to multiple ontology terms and vice versa. [10:47] AmandaVizedom: @Melissa: slide 15: Do you mean that experts cannot articulate the distinctions? Or that the modeling language is not expressive enough to capture it clearly? [10:52] SteveRay: @Amanda: How would you handle the example of the Senator who said "I can't define it but I'll know it when I see it". ...in the context of porn. Quite seriously, it seems that perhaps we sometimes operate with definitions only by example and are unable to define concepts. [10:58] AmandaVizedom: @Steve: I did mean to ask for clarification, not to imply that there are no undefinable concepts. In fact, it is very common for us to use concepts operationally and regularly without there being any set of necessary and sufficient conditions to accompany them. The less technical the concept, I'd argue, the more this is likely to be true. IME, textual definitions that mention and guide against potentially confused concepts can be very useful, though inherently incomplete. [11:00] SteveRay: @Amanda: Agreed. [10:54] AmandaVizedom: @Melissa - slide 18, "true path" violations -- common problem indeed; human brains are not good at keeping that kind of entailment trace. ;-) One technique that can help catch these is the creation of tests that instantiate classes and draw the transitive inferences, presenting the results to the experts, who are more likely to catch it at *that* point if they've created either inconsistencies or undesired implications. [11:10] MelissaHaendel: @amanda: I would love your idea "One technique that can help catch these is the creation of tests that instantiate classes and draw the transitive inferences, presenting the results to the experts, who are more likely to catch it at *that* point if they've created either inconsistencies or undesired implications." This would help biologists enormously. [10:49] TerryLongstreth: @Melissa Slide 17 - Curation status include provenance trace? [10:47] PeterYim: Ed & All: please provide some context wrt to your comment ... otherwise the remarks won't mean much in the proceedings [10:53] HansPolzer: Peter, Melissa: my comments were aimed at the issue of ambiguity and controversy in concept definitions. From a pragmatic/realism approach it may be worthwhile to make more of the context assumptions specific to a given definition, allowing more flexibility in applying a specific ontology or term in an ontology in a given situation/perspective/context. [10:55] PeterYim: @Hans - my comment is strictly aimed at messages people are typing into this chat-room (like, "this is great!" would mean much less than "ref. your slide#12, this is great insight!") since we captured the chat-transcript as part of the proceedings of the session. [10:57] MelissaHaendel: (just lost of her voice connection as she got to the last slide) I'm all done :-) ... Thank you. [10:58] FabianNeuhaus: @ Melisssa, the standards for documentation: is this an OBO Foundry effort? [11:06] MelissaHaendel: @fabian this is a Melissa's team effort ;-). I would like it to be an OBO effort, and I think we are getting traction. [10:58] PeterYim: == EdBarkmeyer presenting ... see: the [ 4-Barkmeyer ] slides [11:00] anonymous morphed into LudgerJansen [11:02] JoelBender: Units! °F! [11:04] LeoObrst: @Ed: would you say that code-lists are controlled vocabularies, and that these map to concepts/classes in a given ontology? Another example are digraphs and trigraphs for countries/states. [11:04] SteveRay: @Ed: I'm interpreting your Code List to be what many call an Enumeration. Yes? [11:05] PeterYim: @EdBarkmeyer and All - some members of the Ontolog community had actually worked on a project "CCT-Representation" back in 2004/2005 to map the ebXML Core Component Types ("CCT") to a First Order Logic ontology (SUMO, MILO, QoS) - see: An Ontological Basis for ebXML Core Component Types - http://ontolog.cim3.net/cgi-bin/wiki.pl?CctRepresentation [11:06] DougFoxvog: Code lists are controlled vocabularies. However, as EdBarkmeyer is saying, the meanings of the codes changes. So, dated uses of the codes is important. [11:11] LeoObrst: @Doug: yes, we see this all the time. There are extreme versioning issues involved in maintaining the mappings from these controlled vocabularies to the ontology/ies. [11:06] anonymous morphed into PavithraKenjige [11:07] DougFoxvog: I note that ISO country codes occasionally change. [11:08] HansPolzer: and countries change and come into being and go away [11:11] DougFoxvog: @Hans: I remember once finding an ISO code for an expired country being re-used over a decade later for a new country. I forget what the example was. [11:14] AmandaVizedom: @EdBarkmeyer - Thanks for this presentation. I've seen this topic persist as a point of conflict on ontology projects -- whether to go the string route or the expression-for-thing route. Least-immediate-effort often pulls people toward the string route, but if there is integration to be done, this is usually a mistake (or at best, pushes the effort down the road, when it will have to be redone). [11:14] HansPolzer: Ref slide 6 in Ed's talk, it would seem to have some temporal context specification - I suppose that's in the administrative records somewhere [11:18] HansPolzer: Looking up the code list is becoming more practical as things become more connected over the internet. But there are also practical constraints on doing these lookups in real time while processing transactions [11:18] DougFoxvog: @EdBarkmeyer: re slides 7 & 8. Having a term mapped to a code & code list is intrinsically a ternary concept. An object that is an instance in a code list can have two mappings from item, one to the code & the other to the code list. However, this only would work if the item maps only to a single code & single code list. [11:30] VictorAgroskin: To do such code models consistently is much easier if your ontology language contains concept of class_of_class. Coding models of this type are very common for systems implemented on ISO 15926. And ISO 15926-6 is partially a mapping of ISO 11179-3. [11:18] == Q & A and Open Discussion ... [11:18] AmandaVizedom: I need to drop off -- thanks to all presenters! Good stuff. [11:19] DougFoxvog: Ciao, Amanda! [11:24] LeoObrst: Gramm Richardson et al gave a talk at STIDS 2012 on an 11179 registry addressing these issues: http://stids.c4i.gmu.edu/papers/STIDSPapers/STIDS2012_T04_RichardsonSchwarz_Constellation.pdf. STIDS 2012: http://stids.c4i.gmu.edu/agenda2012.php. [11:25] PeterYim: I am soliciting help from everyone here: -- for software environment stewards and tool developers, please make sure you participate in the upcoming survey ---and help us get these colleagues of yours to respond to the survey too (they'll be on a wiki, so everyone will know who has or hasn't responded) ... or provide us with pointers so we can reach out to them -- we need ontology evaluation experts and tool developers to participate in the "hackathon" and "clinics" activities [11:26] PeterYim: also ... Join us at the "hackathon" and "clinics" activities - http://ontolog.cim3.net/forum/ontology-summit/2013-02/msg00056.html & http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2013_Hackathon_Clinics [11:35] VictorAgroskin: @Peter, @Patrick, @Valentina - we'll be really happy to see Patrick's team to collaborate in our Clinics activity http://ontolog.cim3.net/forum/ontology-summit/2013-03/msg00011.html ISO 15926 Reference Data Library is a nice combination of taxonomies, quite suitable for exploration by methods described by Patrick. [11:28] PavithraKenjige: About intrinsic and extrinsic evaluation .. those parts that are manufactured by others .. is that considered extrinsic evaluation ?? [11:35] LeoObrst: Thanks, all! Good session. [11:35] JoanneLuciano: Would you repeat Peter's last point? [11:35] PeterYim: join us again, same time next week (Thu 2013.03.14), for OntologySummit2013 session-09: "Building Ontologies to Meet Evaluation Criteria - I" - Co-chairs: MikeBennett & MatthewWest - see developing session page at http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2013_03_14 [11:35] JoelBender: Thank you! [11:35] MelissaHaendel: Thanks everyone. [11:36] KevinSimkins: Thanks to all speakers and host. [11:36] PeterYim: great session ... thanks everyone! [11:36] JoanneLuciano: Very interesting, thanks! [11:36] anonymous morphed into BijanParsia [11:36] AliHashemi: thank you all. [11:36] DavidWhitten: Fascinating insights. Well Done all of you. [11:36] PeterYim: ... session ended == ------