From OntologPSMW

Jump to: navigation, search
[ ]
    (1)
Session Introductory
Duration 1 hour60 minute
3,600 second
0.0417 day
Date/Time September 18 2019 16:00 GMT
9:00am PDT/12:00pm EDT
5:00pm BST/6:00pm CEST
Convener Ken Baclawski

Contents

Ontology Summit 2020 Open Discussion of Knowledge Graphs     (2)

Agenda     (2A)

Conference Call Information     (2B)

Participants     (2C)

Proceedings     (2D)

[12:03] RaviSharma: Ken, good morning from CA     (2D1)

[12:14] David Eddy: re Franz's wish list... that's quite the stack... given most/many organizations are still struggling with software configuration management (SCM), now DevOps.     (2D2)

[12:15] David Eddy: Also... the dream of the "data dictionary" bill-of-materials has had a less than 5% survival rate since mid 1950s.     (2D3)

[12:17] David Eddy: 1960s     (2D4)

[12:21] RaviSharma: John said table databases and also graph databases     (2D5)

[12:22] RaviSharma: John RDF is an example of database     (2D6)

[12:23] RaviSharma: conjunction and relation     (2D7)

[12:24] Gary: a sparse table     (2D8)

[12:25] RaviSharma: logic has two entity FOL     (2D9)

[12:25] David Eddy: STRONG agreement John Sowa's position.     (2D10)

[12:26] Gary: Some background on this topic in Survey of Graph Database Models RENZO ANGLES and CLAUDIO GUTIERREZ see http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.110.1072&rep=rep1&type=pdf     (2D11)

[12:26] RaviSharma: graph database is also based on logic     (2D12)

[12:27] David Eddy: "accepted"... punch cards were accepted as best in the day.     (2D13)

[12:27] David Eddy: people are scared for multiple reasons... scope/size & how to keep updated     (2D14)

[12:27] John Sowa: David, punched cards are the prototype for a table DB.     (2D15)

[12:28] David Eddy: @John... yes... that's why I mention it     (2D16)

[12:28] John Sowa: Every card maps to a row in some table     (2D17)

[12:28] John Sowa: In the olden days, people took courses in punched card design.     (2D18)

[12:29] John Sowa: For every application of punched cards, there is a schema for each type of card.     (2D19)

[12:30] John Sowa: The COBOL data division is the schema for each type of card.     (2D20)

[12:31] Paul Tyson: Stan Ulam said something to the effect of "What makes you so sure human thinking corresponds to mathematical logic?"     (2D21)

[Added Later] KenBaclawski: Here is a more complete quotation "What makes you so sure that mathematical logic corresponds to the way we think? You are suffering from what French call a deformation professionnelle [occupational hazard]. Look at the bridge over there. It was built following logical principles. Suppose that a contradiction were to be found in set theory. Do you honestly believe that the bridge might then fall down?"     (2D22)

[12:32] RaviSharma: bioportal has ontologies     (2D23)

[12:32] John Sowa: There is no such thing as a single ideal ontology.     (2D24)

[12:33] John Sowa: There always has been and always will be an open-ended variety of different ontologies.     (2D25)

[12:35] RaviSharma: one single node called entity and interoperability is fundamental issue - from john     (2D26)

[12:36] RaviSharma: John says CYC 6k micro theories     (2D27)

[12:36] David Eddy: re: interoperability... that is THE thing big, old, complex organizations will pay to do well     (2D28)

[12:36] Gary: I raised the issue of "ontologies" are being built on the fly by looking at entity info "out there." This leads to ontology silos...unless we try to harmonize them as we go of this their way of using ontological engineering.     (2D29)

[12:37] Paul Tyson: Decades of work on "interoperable" ontologies have not led to practical industrial gain (e.g. STEP). In my career I only began to make progress after abandoning this attempt.     (2D30)

[12:37] Gary: So a Q for people building KGs is what ontologies did you use and if you rolled your own, why??     (2D31)

[12:39] David Eddy: @Gary... I would observe that the decades of accumulated operational systems were NOT (consciously) built with ontologies in mind     (2D32)

[12:40] Gary: Graphs may encourage people to think in terms of "neighborhood relations". Models with a basic graph structure provide simple support and visualization of neighborhood relations.     (2D33)

[12:40] Mark Underwood @knowlengr: @Paul - Say more about your STEP complaint? It's been useful from the outside (non-domain expert) as a standard that uses an ontology     (2D34)

[12:40] RaviSharma: graph and relational are different - networks concepts - graph interoperability was discussed     (2D35)

[12:45] Gary: John discussed the infinity of possibilities. These get constrained to actuality by necessities. These may be considered the axioms that make up laws.     (2D36)

[12:55] Mark Underwood @knowlengr: I haven't followed the details of the ontology SDO at W3C, but perhaps what John is saying is that it should be a process standard as opposed to what it seems to be evolving into     (2D37)

[12:56] Ram D. Sriram: John has articulated many points for KGs. I still think we will need to develop a good story for KGs.     (2D38)

[12:58] Mark Underwood @knowlengr: @Ram A candidate "story" (in the agile sense) for KGs is to create a transition process from existing software engineering artifacts that are graph-based to one that supports flavors of more rigorous knowledge tooling, e.g., annotation, ontology-to-ontology connections. I think this is only one of the stories, but it's the one I have in mind from the pragmatist's bias.     (2D39)

[12:59] Ram D. Sriram: @mark: Thanks. I am looking for a good story to sell my management that KGs are the next wave (or may be KGs+NNs).     (2D40)

[13:01] Paul Tyson: @Mark I'm years away from active STEP work. Can comment more next time.     (2D41)

[13:01] Mark Underwood @knowlengr: @Paul, Ack'd     (2D42)

[13:02] John Sowa: Mark, what is pragmatic depends entirely on what software is available TO SUPPORT IT.     (2D43)

[13:04] Paul Tyson: @Mark @John there is lots of open source software to support entire W3C semantic web stack.     (2D44)

[Added Later] KenBaclawski: @John @Paul: I agree. Bipin Indurkhya and I wrote a paper (available at [1]) on the problem of whether Square is a subclass of Rectangle or the other way around or neither. The problem with thinking of Square as being a subclass of Rectangle is that one is viewing the concepts set-theoretically without any behavior. As we stated in our paper: "... the concept square, which is defined to be the class of all squares without any actions on them, is not the same as the concept square in which the objects are allowed to be shrunk or stretched." Similarly, the concept of knowledge graph should be viewed as not just a set-theoretic notion but rather one with a great deal of behavior (i.e., software to support it).     (2D45)

Resources     (2E)

Previous Meetings     (2F)


Next Meetings     (2G)