From OntologPSMW

Jump to: navigation, search
[ ]
    (1)
Session Synthesis
Duration 1 hour
Date/Time 31 Mar 2021 16:00 GMT
9:00am PDT/12:00pm EDT
5:00pm BST/6:00pm CEST
Convener KenBaclawski
Track Synthesis

Contents

Ontologies are a rich and versatile construct. They can be extracted, learned, modularized, interrelated, transformed, analyzed, and harmonized as well as developed in a formal process. This summit will explore the many kinds of ontologies and how they can be manipulated. The goal is to acquaint both current and potential users of ontologies with the possibilities for how ontologies could be used for solving problems.     (2A)

Agenda     (2B)

Conference Call Information     (2C)

Attendees     (2D)

Discussion     (2E)

Requirements for Definitions     (2E1)

[11:59] AlexShkotin: Track B. Definitions.Synthesis:we should continue to study systems of definitions in different ontologies in the direction of total formalization.     (2E1A)

[12:12] ToddSchneider: What are the requirements for 'definition'?     (2E1B)

[12:16] AlexShkotin: @Todd, to be precise, full and contextless.     (2E1C)

[12:19] AlexShkotin: Definitions are for knowledge acquisition and agreement.     (2E1D)

Ontology Summit Track Organization     (2E2)

[12:16] RaviSharma: An excellent overview was presented by Ken to get us started on synthesis session, an excellent job of getting major aspects from each speakers.     (2E2A)

[12:17] RaviSharma: However assumption that we can synthesize these tracks is itself my question?     (2E2B)

[12:17] RaviSharma: Does the inter-track synthesis make a sense logically and practically.     (2E2C)

[12:18] RaviSharma: Logic because are tracks logically connected naturally or are forced to integrate or synthesize?     (2E2D)

[12:19] RaviSharma: Also if these are synthetic as Todd suggested then does the synthesis result in a meaningful end goal or understanding that enhances the theme of Harmonization?     (2E2E)

[12:22] RaviSharma: So my Q is what aspect or track is most relevant for overall ontology harmonization?     (2E2F)

[12:34] RaviSharma: Gary says - these are important part of landscape, little bit of history and how we started, we can integrate them.     (2E2G)

[12:39] RaviSharma: So we can still address at least two parts independently for a while say harmonization and we do later need to address it with Generation as well.     (2E2H)

Definition of Harmonization     (2E3)

[12:21] RaviSharma: Gary described the harmonization of definitions, my Q is whether it leads to harmonization of ontologies that result from harmonized definitions.     (2E3A)

[12:23] RaviSharma: I am not currently addressing generation but asking the question whether harmonization of ontologies is dependent on how similarly these are generated?     (2E3B)

[12:25] Gary Berg-Cross: @Ravi et al. Harmonization, which needs to be defined in and for the communique, needs to occur at each level of the spectrum and within it.     (2E3C)

[12:31] ToddSchneider: harmonize =def. make consistent; produce a pleasing visual combination     (2E3D)

[12:35] ToddSchneider: One way to (attempt to) develop ontologies that provide a greater chance for 'harmonization' (or integration?) is to align with some foundational ontology (in development).     (2E3E)

[12:35] janet singer: Attempt to capture what Ravi said: Does how we generate ontologies impact their harmonization?     (2E3F)

[12:36] ToddSchneider: Janet, definitely.     (2E3G)

[12:41] RaviSharma: excellent from Todd     (2E3H)

[12:43] RaviSharma: Gary addressed that NL and other aspects of harmonization were addressed in his presentation     (2E3I)

[12:43] RaviSharma: Logical way but some things are difficult to harmonized     (2E3J)

[12:44] ToddSchneider: How should 'harmonization' of ontologies be understood? Logical consistency? Ontological Consistency? Implementation consistency? Model equivalence (i.e. the set of models of the ontologies are 'equivalent')?     (2E3K)

[12:46] RaviSharma: Todd - Harmonized implies sing together in harmony similarly work together     (2E3L)

[12:48] RaviSharma: Gary- Onto spectrum in communique address harmonization at different levels also meaning axioms are harmonized at representational level.     (2E3M)

[12:49] RaviSharma: Todd- Set of models for larger than those for CL?     (2E3N)

[12:49] Andrea Westerinen: @Todd, but OWL + SWRL can express this. I think that we need to clarify the distinction.     (2E3O)

[12:49] RaviSharma: Meant based on OWL will be more than those for CL     (2E3P)

[12:51] janet singer: Taking off on Ravi's point re sharpening an organizing question: How do the various dimensions of ontology generation (automatic vs manual, different learners, contexts for communications, conceptualizations and definitions, domain knowledge and formalizations, logics) lead to issues in future harmonization and sustainability?     (2E3Q)

[12:52] RaviSharma: Ken - Are there standards? for ontologies, and for harmonization?     (2E3R)

[12:59] ToddSchneider: Harmonization and reuse?     (2E3S)

[13:00] RaviSharma: Ravi says - let us have notion of heterogeneous ontology creation as one use case and let us study harmonization among such ontologies namely what exists today. Then let us study use case 2 where these are generated by predefined requirements and known technologies, then study their harmonization.     (2E3T)

[13:01] Gary Berg-Cross: I use the semantic gradient/ladder to discuss harmonization. It consist of a range of semantic resources including vocabularies, taxonomies, thesauri, & ontologies arranged by their degree of formal representation. Contributions across the ladder even at the lower end (e.g. vocabularies, taxonomies etc.) are valuable and have their niche - but we must be clear where stuff is, what it's suited for, and what its drawbacks are, perhaps how it was created as well as how it is related to other information and knowledge across the semantic spectrum. A goal of developing semantic resources is that resources on each rung of the ladder can be understood to contextually enhance/feed the others above and below. In a sense they can be "harmonized" with each other. [13:02] BobbinTeegarden: Do rungs of the ladder represent levels of abstraction? Detail?     (2E3U)

[13:03] ToddSchneider: Are harmonization and reuse (operationally) equivalent?     (2E3V)

[13:09] RaviSharma: Gary and Robert Rovetto have discussion on horizontal spectrum, levels of ontologies? Levels assume architecture.     (2E3W)

[13:18] Robert Rovetto: @Ravi...to clarify your restatement: the past discussion (a brief email) was simply that I suggested a spectrum is more precise. Re:levels, that was not in the discussion — rather, just a point I've argued over last few years at different venues, namely that the characterization of an ontology in terms of levels (e.g., lower level) is not precise enough and assumes an certain architecture. Not all ontological systems employ a levels approach. It is one of many ways to conceptualize an ontological system.     (2E3X)

Role of Context     (2E4)

[12:25] RaviSharma: Alex and Todd - context is model in physics with its underlying parameters being used in context of that model?     (2E4A)

[12:34] RaviSharma: role of context for example Gary said.     (2E4B)

[12:36] Mike Bennett: Context is simply a nexus of concepts, that would be in the concept ontology itself.     (2E4C)

Disharmonies     (2E5)

[12:47] Ken Baclawski: If ontologies disagree would explicitly specifying how they disagree be a form of "harmonization"? Should one always try to specify such disagreements?     (2E5A)

[12:50] RaviSharma: Ken and Todd and Gary - Mutually logically inconsistent?     (2E5B)

[12:51] RaviSharma: Ken - Bio Taxonomies explicitly disagree but there are documented?     (2E5C)

[12:51] RaviSharma: Gary - Criteria for life at taxonomy level?     (2E5D)

[12:53] RaviSharma: Todd - we can examine logical inconsistencies in light of harmonizations     (2E5E)

[12:53] janet singer: A taxonomy of ontology pathologies is a good idea Failure Modes and Effects Analysis for ontologies     (2E5F)

[12:53] RaviSharma: Ken - modularizing, different purposes?     (2E5G)

[12:54] Mike Bennett: No ontology is 'right' (or wrong). Each represents someone's conceptualization of the world.     (2E5H)

[12:54] RaviSharma: To Ken - do you mean limiting" to be fit for the purpose" this would be intentional?     (2E5I)

[12:55] Gary Berg-Cross: Historical attempts to standardize/harmonize terms included creating core metadata models and common conceptual models for combining data into a single representation. These however have largely failed to be adopted because of flawed conceptualizations, lack of community agreement, inadequate representation and thus amount to the creation silos. We are trying to address this now with improved methods.     (2E5J)

[12:55] RaviSharma: Janet and Ken - disharmony does not imply not working together.     (2E5K)

[12:56] RaviSharma: The nature of disharmony is real - but we work around them in terms of Statistics and probabilities and yet many times end results make sense, not always?     (2E5L)

[12:58] Gary Berg-Cross: The problems of inconsistent term concepts - These come about of many in-house terminologies which lead to a proliferation of idiosyncratic, and in many cases, arbitrary variation in how scientific terms are defined and inter-related. This leads to significant challenges in finding and integrating relevant earth science information across independently-derived, distributed information sources. The lack of consistent linkage severely hinders progress on investigations that transgress traditional disciplinary boundaries, even though these types of efforts are increasingly needed to address many of the most critical, complex, inter-related of various science research challenges coming with interdisciplinary efforts.     (2E5M)

[13:01] RaviSharma: When you compare the two, you will probably see indeterminacies that are based on how domain expertise enters these generations?     (2E5N)

[13:03] BobbinTeegarden: Ravi are you implying that level of abstraction is pertinent in harmony?     (2E5O)

[13:06] RaviSharma: Bobbin discussed levels of abstraction and common definitional level - it was useful.     (2E5P)

Ontological Commitment     (2E6)

[13:02] RaviSharma: Mike Bennett started discussion on ...?     (2E6A)

[13:03] RaviSharma: Mike please fill in     (2E6B)

[13:03] Mike Bennett: Ontological commitment - is the source of many disharmonies among ontologies. Need to formalize that. Challenge: if there are autogenerated ontologies, how are they to address questions of ontological commitment? That process needs to be managed in some way.     (2E6C)

[13:05] Andrea Westerinen: @Mike does having a human-understandable view (such as related to events) help in regard to the ontological commitments? I would like to explore that with you.     (2E6D)

[13:05] Mike Bennett: @Andrea interesting idea. Humans all have their own ontological commitments - more varied than we expect of each other.     (2E6E)

[13:06] Mike Bennett: @Andrea happy to have a chat about that.     (2E6F)

[13:06] janet singer: I don't see how one can analyze generation and harmonization without exploring what ontological commitment is     (2E6G)

[13:09] ToddSchneider: What is the relation(s) among 'abstraction', 'generalization', 'ontological commitment', 'set of models (a la Tarski)'?     (2E6H)

[13:09] janet singer: Andrea: Yes, having an event perspective goes together with anticipating/understanding the operational effects (intended and untended) of using an ontology or of using a tool with given ontological commitments.     (2E6I)

[13:25] Robert Rovetto: Re: ontological commitments...this is worth exploring. When it comes to a computational ontology, rather than a philosophical one, it is more difficult to determine. Because the latter will, ideally, know to make all assumptions explicit and will ideally have a significant degree of introspection. Since the audience or pool of potential computational ontology developers are from a wider set of backgrounds, knowledge, and understanding, we cannot assume that when a person makes an ontology that they have thought about assumptions they may be making. That is, it is not necessarily the case that a person is making ontological commitments for their computational ontology, in part because they may not. An interview with them can reveal some by asking what the intended meaning of this or that construct in their ontology is. But without knowing this we cannot say all are making commitments. The only sense in which we can say that is if their ontology adopts a particular logic and thus a particular set of inference rules. Then one can say that this or that follows from something else.     (2E6J)

Landscape Issues     (2E7)

[13:04] Gary Berg-Cross: As part of synthesis have we/are we addressed some landscape issues raised?     (2E7A)

  • How are the ontologies expected to be used?     (2E7B)
  • How will changes to the ontology and entailments be rolled out?     (2E7C)
  • How will ontologies be sustained? What does this entail?     (2E7D)
  • What is the architecture of the ontologies and associated constructs (e.g. queries)?     (2E7E)
  • What is the architecture of the information system that will use the ontologies?     (2E7F)
  • How will any transition (of the information system to using ontologies) be accomplished?     (2E7G)

Definition of "Term"     (2E8)

[13:06] Mike Bennett: We need to define the term 'Term'     (2E8A)

[13:07] ToddSchneider: term =def. a word or phrase used to describe a thing or to express a concept, especially in a particular kind of language or branch of study (Will this do? OR do we need to being sequences of symbols?)     (2E8B)

[13:08] RaviSharma: Robert Rovetto - mentioned language - ontology - use of category - difficult to imply a single meaning to a word in a language such assertions may be difficult     (2E8C)

Adjournment     (2E9)

[13:10] Cas Miles: Thank you all, very interesting things to think about!     (2E9A)

[13:10] RaviSharma: Thanks everyone     (2E9B)

Resources     (2F)

Previous Meetings     (2G)


Next Meetings     (2H)