|Date/Time||November 10 2016 17:30 GMT|
|9:30 PST/12:30 EST|
|5:30pm BST/6:30pm CET|
Domain Vocabularies, Ontologies and Semantics- Mini-series Home Page
This is a 2 session Mini-series on "Domain Vocabularies, Ontologies and Semantics" - a Domain Vocabulary-Ontology Dialog.
About the Series:
Description, Goals and Objectives This mini-series is designed to explore the current status and application of semantic science and ontologies to systematize and leverage the already large body of domain definition work on vocabularies and their meaning. Increasingly standardizing the registration and management of domain vocabulary is needed since such terms are used as part of data/metadata documentation. As part of this controlled vocabulary are often used for indexing and retrieval of data resources. However terminologies, driven in part by early efforts, are often arbitrary with little supporting conceptualizations or real standardization. There are some cases with many dozen local standards for data which makes data integration a challenge. Despite community efforts various types of standardized vocabularies are, for the most part, heterogeneous, meaning they:
- are mostly fragmented and disconnected, describing without a foundational grounding.
- This results in domain vocabularies that have the same concept scope; but are represented with different terms, use different formats and formalisms, and are published and stored with alternative access methods. Indeed, since the “vocabularies” are often just defined words, rather than defined concepts with word labels, there is a struggle to reach agreement because various groups have independently develop very similar concepts with their own vocabulary. And as noted in previous Ontolog Forum discussions, distinguishable concepts can be lumped together into a single concept and 2 words many seem to have the same meaning but there are granularity, time scale or agency distinctions that can be made. This creates the need to modify existing vocabularies extensively including refining one concept by adding another subordinate concept or a concept participating in an additional semantic relation.
Past Ontolog Forum sessions have also demonstrated, for example, problems in the Earth Science where many alternative vocabularies compete. As an example, some water quality vocabularies conflate multiple concepts and insert these into a single, compounded term. Thus a term for one type of observation may mix the substance (or taxon) with the medium (e.g. water) observed, along with the procedure used as part of the observation and the units used for measurement. Alternate terms, the lack of clear definitions and poor maintenance of vocabularies makes systematic vocabulary use as well as integration with other vocabularies difficult.
This comes despite decades of intensive work on controlled vocabularies (standardized sets of terms) and now human readable definitions accessible via URIs on the Web, problems remain. While linked data using RDF/S provides some help with representation and syntax there is often no supporting, systematic conceptualization. Some work with leverages classification schemes and thesauri but lack well-defined semantics and structural consistency, which makes matching up concepts that terms represent difficult. Part of the remaining task is then the lingering challenge of clarifying the alternate representation of implicit ideas in people's interpretation to reasonably reflect the types of entities found in reality.
Is there a rationalized “consensus” on how to leverage existing vocabularies as well as existing ontologies? In past Ontolog discussions approaches as different as top-level ontologies, ontology design patterns, RDF graphs, bridge ontologies and reference ontologies have been featured. This mini-series is an effort to update and advance the discussion within the Ontolog Forum Community and beyond to communities like the Research Data Alliance which also has an interest in improved semantics for domain vocabularies.
A key mission of the mini-series is to bring together members of various domains into a meaningful dialog. We anticipate that the sharing of problems and issues, ontological engineering architectures and approaches, and prospective tools, will enable collaborative understanding of the challenges and potential value in the application of ontology and semantics for people currently struggling with domain vocabularies.
Info on Our First session
- 2016_11_10 - Thursday:
This session represents a follow up to some of the things discussed at past Ontolog Forum sessions and the recent RDA Plenary 8 in Denver which included a BoF on Domain Vocabulary Development, Standardization, Registration, Harmonization and Support.
Gary Berg-Cross will provide an introduction to the session and the topic of improved semantics for domain vocabularies drawing from past Ontology Summits on the role of semantics and ontologies in Big Data (2015) and IoT (2016). Introduction Slides
- Mark Fox (University of Toronto) will talk on: An Ontology Design Pattern for Global City Indicators Slides
Abstract: Efforts are underway to introduce international standards for vocabularies and ontologies for Smart Cities. One area of standards development is city indicators for measuring quality of life, sustainability and resilience. This presentation will review: 1) the city indicator standard efforts as embodied in ISO 37120 and ISO 37121; 2) concepts that underly the representation of indicator definitions and data; and 3) a new ISO project JTC 1 N130171 “An Upper Level Ontology for Smart City Indicators”.
- Torsten Hahmann (University of Maine) will talk on: "Domain Reference Ontologies vs. Domain Ontologies: What's the Difference? Lessons from the Water Domain" Slides
Abstract The talk outlines an approach of using a domain-specific reference (or foundational) ontology as a tool for integrating existing semi-formal ontologies within a common domain. We will illustrate this using a reference ontology for the water domain, showing how it has helped formalize, refine and revise an existing semi-formal hydro ontology as a stepping stone towards improved semantic integration within the water domain. We present key characteristics that a domain reference ontology must fulfill and how they compare to what a typical domain or application ontology looks like.
- Boyan Brodaric (Geological Survey of Canada/Natural Resources Canada) will talk on: What's a river? A foundational approach to a domain reference ontology for water Slides
Abstract The role and need for domain reference ontologies is becoming more apparent in ongoing work in hydro ontology design. Such ontologies provide an "upper" domain layer that is grounded in foundational principles and which can thus serve as a bridge to established foundational ontologies. This presentation illustrates the partial development of such a layer for the water domain, with a focus on water bodies such as rivers and drawing inspiration from Hayes' Ontology for Liquids.
Additional Resources: We plan to post a recording of the session here.
Conference Call Details
Date: Thursday, 10-Nov-2016 Start Time: 9:30am PST / 12:30pm Eastern / 6:30pm CEST / 5:30pm GMT / 1730 UTC
ref: World Clock
Expected Call Duration: ~2 hours
- Dial-in and Screen-share (GoToMeeting):
- Please join the on-line meeting (screen share and voice over IP access).
- Use your microphone and speakers (VoIP) - a headset is recommended.
- Or, call in using your telephone on one of the numbers below.
- United Kingdom: +44 (0) 20 3713 5011
- Australia: +61 2 8355 1034
- Austria: +43 7 2088 0716
- Belgium: +32 (0) 28 93 7002
- Canada: +1 (647) 497-9372
- Denmark: +45 69 91 84 58
- Finland: +358 (0) 923 17 0556
- France: +33 (0) 170 950 590
- Germany: +49 (0) 692 5736 7206
- Ireland: +353 (0) 19 030 053
- Italy: +39 0 699 26 68 65
- Netherlands: +31 (0) 208 080 759
- New Zealand: +64 9 974 9579
- Norway: +47 21 04 30 59
- Spain: +34 931 76 1534
- Sweden: +46 (0) 852 500 691
- Switzerland: +41 (0) 435 0026 89
- United States: +1 (669) 224-3319
- Access Code: 691-770-629
- Meeting ID: 691-770-629
- Audio PIN: Shown after joining the meeting
- or just press #
- Online Meetings Made Easy®
- Not at your computer? Click the link to join this meeting from your iPhone®, iPad®, Android® or Windows Phone® device via the GoToMeeting app.
In-session chat-room url:
See transcript below http://webconf.soaphub.org/conf/room/voc_20161110
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").
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.
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) firstname.lastname@example.org ... Handy for mobile devices!
Chat transcript from room: voc_20161110 2016-11-10 GMT-08:00 (UK time I think) List of attendees: Alberto, Alberto1, AlexShkotin, Ant, Boyan Brodaric, Chuck Vardeman, CoryCasanave, DaliaVaranka, DaliaVaranka1, Dennis Pierson, Donna Fritzsche, Jack Hodges, Josh Lieberman, Kenneth Baclawski, Kim Tryka, Mark Fox, Michael Denny, MikeBennett, Ouida Meier, PaulTyson, SteveRay, TerryLongstreth, ToddSchneider, Torsten Hahmann, anonymous, anonymous1, anonymous2, brandon, brandon whitehead, gary berg-cross,
[09:39] AlexShkotin: there is a chat inside GoToMeeting:-)
[09:41] AlexShkotin: I know
[09:48] Jack Hodges: schema.org should include BaseURIs or the models presented have no verifiable/vettable provenance [09:54] Chuck Vardeman: Peter Patel-Schneider also did an analysis of schema.org semantics.
[09:54] Chuck Vardeman: http://iswc2014.semanticweb.org/raw.githubusercontent.com/lidingpku/iswc2014/master/paper/87960257-analyzing-schemaorg.pdf?raw=true [09:56] MikeBennett: An interesting feature of schema.org that was explained to me when we integrated FIBO based terms, is that schema is explicitly polysemic (their term) i.e. a credit card can be both a product and a facility and so on.
[09:57] AlexShkotin: it would be nice to have one bigger slide on screen
[10:02] gary berg-cross: Todd Schneider provid some discussion of the vocabulary-ontology issue in an ealier email: So 'vocabularies' provide a good starting point for developing an ontology for that domain. They provide some idea of the entities and relations that are 'present' or relevant to the participants and their operational activities. There may also be definitions. But vocabularies in many domains usually have embedded implicit assumptions about the domain or contexts that occur in that domain (e.g., frames of reference). My use (and interpretation of) ontologies is that they are designed and developed for use in human developed systems to provide an explicit constraint(s) on the interpretation of identifiers (almost always natural language terms). In this context an ontology can be used by (current) computers with the expectation that any results produced will not be too dissimilar from results produced by a human (following the same process/algorithm). Ontologies are engineered artifacts developed to meet some needs. But those 'needs' [that an ontology is supposed to provide for] have scope. As such they usually embed aspects of those needs and their scope (in different ways). So in the development of an ontology for a particular purpose or, perhaps a range of similar purposes, starting from an existing vocabulary and existing definition is efficient way to start. Unfortunately, many people stop there or shortly thereafter relying on assumed interpretations (of natural language terms), i.e., almost direct use of terms from a (domain) vocabulary is made. Making direct use of a vocabulary for the basis of an ontology will almost always produce something of limited utility (which may be sufficient:). With the added advantage that 'users' in the domain (including any other 'developers' involved) will feel comfortable and not need to learn additional notions. In working with OWL I've adopted the practice of using OWL annotation 'label' (or add one) to associate a 'display' label with an entity (class or relation). Or just add a data property for the display label. Such constructs can also be used to support multiple (natural) languages. This is similar to the Model-View-Controller pattern from software (in which 'model' refers to a data model). The 'model' is the ontology; the view is comprised of the display labels (i.e., the vocabulary terms) an end user views; the controller is the implementation the employs the ontology. [10:05] Donna Fritzsche: I like that description Todd! [10:10] Donna Fritzsche: Can someone explain how the term pattern is being used here?
[10:10] Donna Fritzsche: It looks more like schema or rule
[10:11] gary berg-cross: Mark, it would be interesting to hear what others in this domain think about this foundational work. Do they recognize it's value and what to they say about the level of effort needed?
[10:22] Mark Fox: In working with the City Anatomy initiative (Barcelona), RDA, and ISO, there is a continuum of understanding. Within ISO, many of those who specify indicators have little understanding of the data representation side of the world. But those on the JTC side of ISO have a growing understanding. Certainly, vocabularies are in vogue, but more are beginning the understand the importance of semantics and ontologies.
[10:24] Mark Fox: Within the world of cities, the same continuum of understanding exists, with those who understand the importance of ontologies being outliers. The main challenge is cities will not understand the importance of ontologies until there is an important application that requires ontologies.
[10:25] MikeBennett: What we're seeing in finance is that the business side is seeing the need for computationally independent representations of concepts, but only know about Vocabulary as a way of doing that. Meanwhile ontology is regarded as another technology matter and therefore not relevant to that problem - whereas of course a reference ontology is exactly what they need at the business level.
[10:26] ToddSchneider: Torsten, are there more explicit requirements for your reference ontology?
[10:29] Mark Fox: I am working with the City of Toronto on the problem of tagging all of their documents, which there are millions, with keywords.
They have a controlled vocabulary which is published on their open data web site, but it is semantically a mess. This is a problem that all cities have and there are vendors that have tagging tools using controlled vocabularies. We are even seeing SKOS being used in them. The project we are working on is developing an ontology of city subjects and services. Problem is cities reflect the variety of the world at large, which means it covers a lot! Have to focus on a subarea. [10:29] gary berg-cross: Something on OGC'S Groundater ML http://ngwd-bdnes.cits.rncan.gc.ca/service/api_ngwds/en/gwml.html
[10:30] MikeBennett: @Mark if they are using SKOS you can introduce an arrangement whereby SKOS is generated from a set of OWL ontologies, providing a relatively business friendly vocabulary but knowing it's consistent underneath.
[10:31] Mark Fox: @Mike - good to know. Thanks.
[10:32] brandon whitehead: @Torsten et al., using the notion of a "container", or "containment" is an interesting approach. How might you reconcile surface water--groundwater interaction?
[10:35] gary berg-cross: Torsten/Boyan You certainly put us on an intuitively sensible model path that is far from a "flat" knowledge space.
[10:37] SteveRay: @Torsten: How did you test the consistency of your models? @Torsten: Never mind - I see that you mention you used theorem provers.
[10:40] Josh Lieberman: It will certainly be interesting to apply HyFO to the new HY_Features model and NHDHYF ontology. There is an aspect of the domain that may still need to be included, namely that there is useful data and useful work that demonstrates connectedness of water but does not completely characterize a physical hydrology domain such as a catchment.
[10:41] Josh Lieberman: Domain reference ontologies have a pretty large affinity to ontology design patterns. Is there a clear difference?
[10:43] Josh Lieberman: Connecting surface and ground water will be an important next step to take, however the conceptualizations of such interfaces are not as mature as either domain above or below them.
[10:44] Mark Fox: @Josh The design pattern I showed uses reference ontologies to construct the pattern. So the pattern is basically a ratio of population measures. The reference ontologies include: measurement, time, populations, etc.
[10:45] ToddSchneider: Considering the usual uses of the term 'foundation' (i.e., at the bottom, an underlying basis or principle for something) and what it is intended to convey in the domain of ontology, I suggest we drop use of the term 'upper' in referring to a foundational ontology. And change diagrams to present a foundational ontology at the bottom of the diagram.
[10:46] Torsten Hahmann: @Brandon: there are two aspects: one are the things that are common to surface and groundwater like the existence of some container (or support) object, a water object and its phyiscal relation (one being inside the other). The second are explicit interface objects, such as springs & wells & infiltration. If we have those formalized, we can actually do the kind of integrated reasoning (e.g. "water contaminants in surface runoff or surface water gets to what aquifers"). We did formalize springs and wells from GWML2, but would be curious to see how it fits other definitions.
[10:46] Ant: to repeat @brandon - I'm interested in how near surface interactions and 'soil water' in general could be extended in this framework.
[10:48] Josh Lieberman: There are some easier steps to take, such as infiltration zones and stream bed leakance, but it gets more complicated from there.
[10:49] Torsten Hahmann: @Steve: we could only test the internal consistency of the ontology and some extra sanity constraints (such as that concepts can actually be instantiated without causing an inconsistency). But we are working on methods that can test the consistency with small data sets, which in our experience theorem provers don't handle well.
[10:49] CoryCasanave: In the threat/risk model we are doing (threatrisk.org), we have had to do the same thing- deal with a lot of common concepts that risk and threat relate to or build on. Having a better library of concepts that can compose and provide linking points between deomains would have value
[10:50] Torsten Hahmann: @Todd: that is a very good point. The diagrams may make more sense with the foundation at the bottom.
[10:51] Josh Lieberman: Not sure that is consistent with the lofty goals of many upper ontologies, but to my mind it makes more sense.
[10:54] ToddSchneider: Torsten, are there more explicit requirements for your reference ontology?
[10:55] Ant: I'm not sure I would class an estuary mouth as a fiat boundary. Perhaps to a geologist it may be but not to a hydrographer. Brandon Bennett has touched on some of these issues: http://ijcai.org/Proceedings/05/Papers/1187.pdf
[10:56] Torsten Hahmann: @Josh: can you clarify your earlier comment regarding catchment. Are you saying these things are another step beyond the surface-subsurface water interface? I do agree that the interface concepts will probably be the trickiest part, simply we don't have a solid understanding of what they really are.
[10:56] Mark Fox: @Todd - there are two diagrams: 1) a taxonomy where the "foundational" classes are at the top or upper levels, and 2) an ontology module diagram where the foundational ontologies are at the bottom. So the distinction is "foundational ontologies" vs "upper level concepts".
[10:59] MikeBennett: @Boyan is there any reason you are showing a monohierarchical taxonomy with e.g. UncontainedSupported as a sibling of UncontainedUnsupprted rather than polyhierarchy where Contained and Supported would be separate parents?
[10:59] ouida morphed into Ouida Meier
[10:59] Josh Lieberman: It seems that one role for one of these domain ontologies is to conceptualize the physical or scientific principles that apply to a domain, such as containment. In the hydro domain, many concepts derive from the conservation of water mass or energy.
[11:00] ToddSchneider: Mark, what you described is just a particular convention. The question is creating any diagram (or other information artifact) is what the creator/author would like to convey. 'Upper level concepts' are those that underlie other concepts. Hence they're at the bottom:)
[11:01] Torsten Hahmann: @Todd: the real requirement was that concepts in any of the domain ontologies logically extend only concepts from the domain reference ontologies. In that respect, it cleanly separates (sub-)domain concepts from foundational concepts. That might be an ideal, though.
[11:01] Josh Lieberman: A tricky part of then constructing a domain ontology is that it may need to accommodate information for which the physical principles may not be very clear. For example, we may be able to determine that a particular stream reach is "gaining" from groundwater or "losing" without having determined the precise locations or mechanisms for that exchange of water.
[11:01] ToddSchneider: Have to go. Thank you.
[11:02] Ouida Meier: If Domain Reference Ontologies (DRO) are needed to reconcile and unify multiple Domain Ontologies (DO) but DROs are not useful in applications, how does useful information from the DRO work for improved vocabularies and ontologies make their way back down to application ontologies? How do we apply the results of this work consistently to practical data collection, storage, interoperability, retrieval, analysis, and modeling?
[11:04] brandon whitehead: @torsten, the reason I asked about the interaction is because of the use of _containment_. Modelling surface water and groundwater as disjoint (through containment) is a simplification, and a useful one in many applications, but if there were a need to model a system to trace a pollutant (for example) how might that be incorporated in the model?
[11:05] brandon whitehead: I'm not suggesting there is an answer, just wondering where you are in your thought process.
[11:05] gary berg-cross: Please put your Qs in the chat as a record.
[11:06] gary berg-cross: My Q was are others using domain ref ontologies.
[11:06] brandon whitehead: @Josh, I'm not sure the issue is with the physics, or physical principles, but with the uncertainty with rates and location.
[11:06] Josh Lieberman: @Mark Fox - so it seems you are saying that domain reference ontologies and ontology patterns are almost orthogonal, but using terms from reference ontologies can increase the generality of patterns.
[11:09] Josh Lieberman: @brandon - we often do not have enough data and cannot measure in sufficient detail to determine e.g. the physics, but still derive useful information that we'd like to represent with ontologies.
[11:09] brandon whitehead: @Torsten, also, I should mention that I did not have a chance to review the slides ahead of time. Perhaps it will be clearer when I review the diagrams in more detail. :)
[11:12] brandon whitehead: @Josh - agreed; that's Earth science. :)
[11:12] gary berg-cross: I think of ODPs as potentiall independent of domain ref ontologies (DROs) rather than the 2 being orthogonal. ODPs can be used to support varioua DROs as a best pracice.
[11:13] gary berg-cross: Torsten We might call this a critical mass of ontologies.
[11:14] Josh Lieberman: It seems to me that ontologies supporting computation are still weak.
[11:14] Torsten Hahmann: I like that term!
[11:16] Torsten Hahmann: @Brandon: can you bring up your question/remark in the discussion? I'm not sure I fully understand
[11:16] gary berg-cross: Q on getting better definitions from domain people. Mark's point is that domain people need to be motivated.
[11:17] MikeBennett: The need to use imagination is a challenge both with business domain experts and with IT folks in my experience.
[11:18] gary berg-cross: How about Brandon's Q on: Modelling surface water and groundwater as disjoint (through containment) is a simplification, and a useful one in many applications, but if there were a need to model a system to trace a pollutant (for example) how might that be incorporated in the model?
[11:20] Ouida Meier: If a core group of domain scientists needs to take a (modest) collection of (mostly elderly) vocabularies and ontologies and try to map or aggregate them into a more coherent collective (driven by data needs), how would you recommend proceeding? Including software tools at the more-accessible less-coding end of the spectrum?
[11:20] gary berg-cross: Brandon"s early Q This falls into the area of Flow, internal etc. Torsten/Boyan ha started looking at this. A pollution plume is contained, which gives us a starting point for understanding but we need more detail.
[11:20] brandon whitehead: Thanks for that clarification @Torsten and @Boyan.
[11:21] MikeBennett: @Ouida I would use a graphical tool like Cameo Conceptual Modeler to map out the concepts in front of the business audience.
[11:22] brandon whitehead: I need to sign off. Thanks everyone!
[11:22] gary berg-cross: @Josh: can you clarify your earlier comment regarding catchment. Are you saying these things are another step beyond the surface-subsurface water interface? I do agree that the interface concepts will probably be the trickiest part, simply we don't have a solid understanding of what they really are.
[11:23] gary berg-cross: @MikeBennett This is for the coral reef research domain - extremely interdisciplinary....
[11:24] MikeBennett: @Ouida there's an art to faciliating that sort of session - starting with basic explanation of the notation (set theory etc. i.e. in English though the underlying logic is FOL or OWL)
[11:26] gary berg-cross: Ouida's original Q " If a core group of domain scientists needs to take a (modest) collection of (mostly elderly) vocabularies and ontologies and try to map or aggregate them into a more coherent collective (driven by data needs), how would you recommend proceeding? Including software tools at the more-accessible less-coding end of the spectrum?"
[11:31] Donna Fritzsche: Gary - what are the details on the boot camp that Torsten mentioned? Gary see http://vocamp.org/wiki/GeoVoCampDC2016
[11:33] Josh Lieberman: Thanks and bye. [11:33] Donna Fritzsche: Thanks all!
- Gary berg-cross,
- Boyan Brodaric,
- Chuck Vardeman,
- Dennis Pierson,
- Donna Fritzsche,
- Jack Hodges,
- Josh Lieberman,
- Kenneth Baclawski,
- Kim Tryka,
- Mark Fox,
- Michael Denny,
- Ouida Meier,
- Torsten Hahmann,
- anonymous, anonymous1, anonymous2,
- brandon whitehead,
Key Date(s) to Note:
- A 2nd session is planned for 2016_11_17 - Thursday
See the session site for more:
We have 4 speakers:
- Simon Scheider (Human Geography and Spatial Planning, Universiteit Utrecht) on the challenges he's found in ontological prerequisites for meaningful spatio-temporal analysis (maps, statistics).
- Olivier Bodenrieder (NIH) will speak on the vocabaulary experience with SNOMED CT.
- Mike Bennett (Ontolog) Topic from FIBO Business Vocabulary
- Cory Casanave (Model Driven) Cross domain sharing and federation of threat and risk information, an application of Semantic Modeling for Information Federation) (SMIF a.k.a. SIMF)
Plan & Schedule
Virtual Panel Sessions planned for this Mini-Series:
- Session-1: Thu 2016_11_10 - Chair: GaryBergCross
- Session-2: Thu 2016_11_17 - Chair: GaryBergCross
Our 2nd session is Nov. 17th.
- Simon Scheider (Human Geography and Spatial Planning, Universiteit Utrecht) on the challenges he's found in ontological prerequisites for meaningful spatio-temporal analysis (maps, statistics).
- Olivier Bodenrieder (NIH) will speak on the Vocabulary experience with SNOMED CT.
- Mike Bennett (Ontolog) Topic from FIBO Business Vocabulary
- Cory Casanave (Model Driven) Cross domain sharing and federation of threat and risk information, an application of Semantic Modeling for Information Federation (SMIF a.k.a. SIMF)