ppy/oor-categories_chat-transcript_unedited_20131008a.txt ------ Chat transcript from room: oor_20131008 2013-10-08 GMT-08:00 [PDT] ------ [6:57] PeterYim1: Welcome to the = OOR Development Session: "Meta-Ontology for Ontology Categories" - Tue 2013_10_08 = Session Topic: Meta-Ontology for Ontology Categories: Developing a consensus for categorization of ontologies in an (Open) Ontology Repository instance Session Co-chairs: Professor Dr. TillMossakowski (U of Magdeburg) & Professor MichaelGruninger (U of Toronto) Briefings: * Dr. AleksandraSojic (U of Bremen) - "Meta-Ontology for Ontology Categories" - opening brief * Professor MichaelGruninger (U of Toronto) - "Ontology Categories in COLORE" Session page: http://ontolog.cim3.net/cgi-bin/wiki.pl?OOR/ConferenceCall_2013_10_08 . == Proceedings == . [6:59] anonymous morphed into AleksandraSojic [7:03] PeterYim1 morphed into PeterYim [7:04] anonymous morphed into JulienCorman [7:05] Daniel Couto Vale: I cannot join the conference. Something wrong is happening to the join-conference account. [7:06] anonymous morphed into TorstenHahmann [7:06] ChristophLange: The account might appear offline in Skype. I was able to dial in nevertheless. [7:07] PeterYim: @Aleksandra, Till and anyone else having trouble connecting - Please use skype-out and dial the telephone number instead (of the "joinconference" user) of you can [7:07] Daniel Couto Vale: For me, it appears offline and the only option is to "Call Phone". There is no "Call" as usual. [7:08] ToddSchneider: Who is ever typing please mute your microphone. [7:09] TorstenHahmann: Hi - I got to call in with skype by using "call" in the menu instead of the button. [7:09] ChristophLange: I muted myself locally [7:10] MariaKeet: I keep on being disconnected (but that's also for the chatroom, so its not due to skype) [7:11] anonymous morphed into PatCassidy [7:16] ChristophLange: Summary: I don't currently have time to follow all OOR activities full-time. But my connection to Till is reliable. Whenever something needs to be done where I can help (such as the stuff I prepared for today), he will let me know. [7:16] Daniel Couto Vale: I'm in. [7:16] ChristophLange: (Sorry, didn't intend to post twice. Software issue?) [7:20] ChristophLange: Till will be with us in a few minutes. [7:20] MariaKeet: sorry, but the network isn't handling the voice well. I'd like to participate a next time, and in the meantime can provide input via email w.r.t. ROMULUS and our experiences with the metadata. [7:21] PeterYim: sorry about this, Maria ... [7:21] PeterYim: == MichaelGruninger starts the session ... [7:22] anonymous morphed into FrancescaQuattri [7:24] Daniel Couto Vale: Please, where is the link? [7:25] ChristophLange: Slides are at http://ontolog.cim3.net/cgi-bin/wiki.pl?OOR/ConferenceCall_2013_10_08#nid3YFE [7:25] List of members: AleksandraSojic, ChristophLange, Daniel Couto Vale, FrancescaQuattri, HensonGraves, JulienCorman, MichaelGruninger, OliverKutz, PatCassidy, PeterYim, RayMartin, TillMossakowski, ToddSchneider, TorstenHahmann, vnc2 [7:26] PeterYim: == AleksandraSojic presenting the opening brief ... [7:29] PeterYim: @Aleksandra - slide#3: what do you mean by "OOR Independent" & "OOR Aligned" [7:33] TillMossakowski: I think "OOR independent" means "sub-ontologies developed by specific communtites", while "OOR Aligned" means "sub-ontologies (typically the upper part) developed centrally by OOR" [7:33] HensonGraves: do you include standard metadata attributes such as where it came from, who did it, when was it done, etc. [7:36] PeterYim: @Aleksandra et al. - are most ontology repository categorization implementations hierarchical now, or do some of them support the assignment of multiple categories to an ontology (multiple inheritance allow in the sturcture)? [7:37] PeterYim: ^allowed [7:40] MichaelGruninger: Actually, COLORE is restricted to Common Logic ontologies -- I think that OntoHub is the only effort that supports multiple ontology languages [7:40] ToddSchneider: Are we clear on the intent of providing a categorization for OOR (or other ontology repositories)? [7:40] ToddSchneider: Is there a focus on 'findability' or 'searcj [7:41] ToddSchneider: Typo correction - Is there a focus on 'findability' or 'search'? [7:42] ToddSchneider: Following up on Peter's comment, would it be better to provide a lattice view? [7:45] TillMossakowski: Aleksandra's domain fields ontology involves multiple inheritance at a few places [7:46] ToddSchneider: Till, how is multiple inheritance presented (i.e., visualized) to a user? [7:47] anonymous1 morphed into RobertRovetto [7:48] TillMossakowski: basically as in Protégé: classes appear multiple times in the hierarchy [7:49] TorstenHahmann: With COLORE we are moving in the direction of multiple classifications ("views") in addition to the main hierarchies (which are based on the set of primitive concepts & relations). E.g. based on whether they can be reduced to others (core vs. non-core) or whether they are generic or domain ontologies, and what larger ontologies they are part of (e.g. PSL or DOLCE). I don't think displaying it as a lattice makes much sense. [7:49] ToddSchneider: Till, understood (about Protege), but what would be presented on a UI (for the repository)? [7:49] ToddSchneider: What is the problem that needs to be solved? [7:56] ChristophLange10: Seems that not everyone can hear me [7:56] ChristophLange10: Should I type into the chat? [7:56] TillMossakowski: yes [7:56] ChristophLange10: OK [7:56] TillMossakowski: Todd, in the UI we present it in a similar way as Protégé does. [7:56] ChristophLange10: The main thing about LoLa is that it is independent from concrete ontology languages. E.g. OMV and the BioPortal/Protege ontologies have a lot of OWL-specific things hard-coded (classes, subproperty axioms, etc.) [7:57] ChristophLange10: LoLa as an ontology abstracts from that, but as an _instance_ of the LoLa ontology we also maintain a _registry_ of concrete ontology languages. [7:57] ChristophLange10: The Lo/La (Logic/Language) level of LoLa can be seen here: http://www.slideshare.net/langec/linked-open-geodata-and-the-distributed-ontology-language-a-perfect-match/26 [7:58] ChristophLange10: The inner-ontology level can be seen here: http://www.slideshare.net/langec/the-distributed-ontology-language-dol-use-cases-syntax-and-extensibility/10. What we show here is not exactly an excerpt from the LoLa ontology, but from the Ontohub database schema, which is modelled after LoLa. (One minor thing is obsolete: we now call "entity" "symbol".) [7:58] PeterYim: == MichaelGruninger presenting "Ontology Categories in COLORE" ... [8:00] ChristophLange10: Yesterday I reviewed the OMV, BioPortal, Protege, etc. ontologies, which Aleksandra has imported into the current draft of the "Ontohub Metadata Ontology", and came up with some concrete ideas for aliging LoLa to the former ones: https://github.com/ontohub/OOR_Ontohub_API/issues (most of the issues so far deal with this; maybe I should assign further tags to them to keep alignment tasks separate from other tasks). [8:01] ChristophLange10: The question from an Ontohub perspective is: We could do without OMV etc.; we could simply "steal" the best ideas from OMV and incorporate them into LoLa. However if we agree that OMV etc. will be of interest to other OOR repositories (maybe particularly the OWL-specific ones), we should rather aim at backwards compatibility and thus at an alignment. [8:02] MichaelGruninger: http://stl.mie.utoronto.ca/colore/org.html [8:02] ChristophLange10: For Ontohub's core purpose (managing ontologies across languages/logics), LoLa is more precise in any case. But LoLa is currently missing metadata/repository/maintenance-related vocabulary, which does exist in OMV etc. [8:02] ChristophLange10: That's it for now about LoLa :-) [8:03] ChristophLange10: Oh, the full background about LoLa is available here: https://svn-agbkb.informatik.uni-bremen.de/sfb-repos/Projects/I1_OntoSpace/papers/TKE2012/DOL-ontograph-layers-TKE.pdf [8:04] ChristophLange10: sorry, wrong link. I meant http://www.informatik.uni-bremen.de/~till/papers/womo2012.pdf [8:05] PatCassidy: Mike: re slide 2 -- in addition to those two characteristics, it could be important to specify *how* an ontology i used, i.e. is it used in some application(s), and if so, which? [8:06] PeterYim: == open discussion ... [8:08] ChristophLange10: If there are any questions about SKOS, I'll be happy to answer them. For now we have a GitHub issue about this: [8:09] ChristophLange10: https://github.com/ontohub/OOR_Ontohub_API/issues/6 - in a nutshell: I think 1. SKOS is semantically more appropriate for domain categories. 2. a lot of category schemes (e.g. librarians') exist in SKOS already, accessible on the Web, and can be linked to. [8:10] Daniel Couto Vale: I would like to comment on search and findability [8:12] TillMossakowski: Peter, you can assign multiple categories in Ontohub [8:16] TillMossakowski: where can I find more about the difference between search and findability? [8:18] PeterYim: @Todd - how about "discoverability" ... that too, I think (or is that synonimous with one of those two - "search" and "findability" [8:21] PeterYim: looks like the topic today has two prongs - should we be dealing with the two issues - "metadata" and "UI" - separately [8:21] Daniel Couto Vale: I would like to place the question again about the function of the categorisation. [8:21] ChristophLange: Sorry, just noticed the link I posted at 15:58 (in my timezone, i.e. 23 min. ago) doesn't work with the dot at the end. Here it's fixed: http://www.slideshare.net/langec/the-distributed-ontology-language-dol-use-cases-syntax-and-extensibility/10 [8:22] ToddSchneider: The enterprise search crowd views discoverability as an aspect of search, a by-product. [8:22] ChristophLange: In my review (see GitHub issues) I found nothing really in OMV that I _disagree_ with. Just lots of aspects where OMV is insufficient. [8:23] FrancescaQuattri: the topic is highly interesting. Maybe the best way to find out the difficulties that users may find in searching ontologies might be solved by asking the users themselves. I know that some unis here run surveys online to work on more user-friendly search systems. questionnaires are also run via crowdsourcing. [8:23] ChristophLange: @TillMossakowski: good point about OMV restricting dual licenses. My view was that on a higher level I found nothing to disagree with; but certainly there might be finer details that we don't accept. [8:23] FrancescaQuattri: our community is quite large, maybe we can run a survey of the major encountered difficulties [8:26] RayMartin: For many topics there are multiple dimensions or viewpoints. Take the topic of life safety - one could develop an ontology of the fundamentals, a scientific perspective. Others may need an engineering perspective. and yet others would utilize a business ontology. i know there are times when i would like to know this type of categorization prior to delving deeply into a given ontology. [8:29] MichaelGruninger: @Peter -- the two issues are merged if we are developing a UI that is driven by metadata [8:29] PeterYim: sounds great, Michael! [8:30] FrancescaQuattri: @Ray. exactly. Alexandra made a good point with Arabic. Let's say one is not interested in all the categories presented in the slides, but let's say in specific ones (even in relation with other languages). the width and depth of the ontology might change considerably. [8:32] ToddSchneider: How could 'tags' be coupled to a 'better' classification? [8:33] Daniel Couto Vale: about keyword tagging [8:33] ToddSchneider: How to go. Thank you. [8:33] ChristophLange: SKOS provides means for connecting keywords to a topic hierarchy [8:33] MichaelGruninger: @Torsten -- is this a meta-ontology or a folksonomy for ontology description [8:34] TorstenHahmann: more of a folksonomy - let it grow through how people annotate ontologies [8:34] ChristophLange: http://www.w3.org/TR/2009/REC-skos-reference-20090818/#labels (notice, e.g., skos:hiddenLabel - something that's never displayed, but accepted as a query string) [8:35] FrancescaQuattri: @Torsten: it might be also interesting to let it grow through what users are looking after/search [8:36] FrancescaQuattri: @Daniel: basically what Daniel is saying. Using frequency [8:37] TorstenHahmann: @Francesca: I totally agree - not just the ontology owner/devleoper should be able to tag, but also all users; they provide probably better tags (how they found it) [8:38] FrancescaQuattri: @Tosten: agreed. Also thinking that developers might estimate a tag/category, users relate the choice -even semantic one for the tag- to their need. [8:38] ChristophLange: Let me type. I think SKOS answers many of our questions. [8:38] ChristophLange: It allows for combining a formal category hierarchy with informal labels [8:39] ChristophLange: As soon as we represent "folksonomy tags" with URIs, we can link them to an existing SKOS scheme [8:45] Daniel Couto Vale: Documenting my comment: the kind of user interaction that we want to allow is dependent on whether we have a folksonomy or a controlled taxonomy. On the one hand, because users tend to be unsure about the tags that would promote their content or be confident in the wrong direction, suggesting the most commonly searched and used tags to the annotator and suggesting the existent tags to the searcher might be a necessary convergence step if we go for an open set of tags. On the other hand, not offering the freedom of choosing any tag might allow us to offer other kinds of user face interaction which are not text fields. [8:41] FrancescaQuattri: hey Everyone, thanks, great talk! [8:41] ChristophLange: Thanks, bye! [8:42] PeterYim: no OOR session next Tue 2013.10.15 - there will be a Common Logic (CLv2) meeting that day [8:42] PeterYim: no OOR session on Tue 2013.10.22 either, ISWC will be in session that week [8:43] PeterYim: @Michael, Todd, Ken, Till, et al. - let's work on the details about the program for the Tue 2013.10.29 session offline [8:44] PeterYim: -- session ended: 8:41am PDT -- [8:45] List of attendees: AleksandraSojic, AleksandraSojic1, ChristophLange, ChristophLange1, ChristophLange10, ChristophLange2, ChristophLange3, ChristophLange4, ChristophLange5, ChristophLange6, ChristophLange7, ChristophLange8, ChristophLange9, Daniel Couto Vale, FrancescaQuattri, HensonGraves, JulienCorman, MariaKeet, MichaelGruninger, OliverKutz, PatCassidy, PeterYim, PeterYim1, RayMartin, RobertRovetto, TillMossakowski, ToddSchneider, TorstenHahmann, anonymous, anonymous1, vnc2 ------