From OntologPSMW

Revision as of 00:10, 6 February 2019 by KennethBaclawski (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
[ ]
    (1)
Session Narrative
Duration 1.5 hour90 minute
5,400 second
0.0625 day
Date/Time Jan 30 2019 17:00 GMT
9:00am PST/12:00pm EST
5:00pm GMT/6:00pm CET
Convener Donna Fritzsche and Mark Underwood

Contents

Ontology Summit 2019 Narrative Theory and Explanation Session 1     (2)

“Narrative theory starts from the assumption that narrative is a basic human strategy for coming to terms with fundamental elements of our experience, such as time, process, and change, and it proceeds from this assumption to study the distinctive nature of narrative and its various structures, elements, uses, and effects.” (The Project Narrative)     (2A)

In this track, we start with a brief discussion of narrative theory and continue with the introduction of related efforts in the ontology, logic and cultural domains, Our goal is to discuss how narrative and narrative constructs can inform and effect explanations and the ontologies which support the explanatory process. There is time allocated for group discussion.     (2B)

Agenda     (2C)

Conference Call Information     (2D)

Attendees     (2E)

Proceedings     (2F)

[11:57] Mark Underwood:     (2F1)

[11:58] KenBaclawski:     (2F5)

12:00pm - 12:10pm Donna Fritzsche: Introduction to Narrative     (2F6)

12:10pm - 1:00pm Ken Baclawski: Proof as Explanation and Narrative Slides     (2F7)

1:00pm - 1:10pm Mark Underwood: Bag of Verses: Frameworks for Narration from Cognitive Psychology Slides     (2F8)

1:10pm - 1:30pm Discussion Time     (2F9)

[12:11] RaviSharma: Ken - the narrative around - before during and after the logical mathematical proof - are they not context + Math / logic?     (2F10)

[12:11] ToddSchneider: What constitutes an explanation? To what extent is 'belief' required (of a context or assumptions)?     (2F11)

[12:12] Mark Underwood: Gary... betting that RDA looked into that use math lit use case?     (2F12)

[12:14] John Sowa: Todd, an explanation is an answer to a question, such as "what does that mean?" or just "why?" or "how?"     (2F13)

[12:14] John Sowa: A proof is a special purpose explanation.     (2F14)

[12:16] RaviSharma: Ken - What happens if the Collection implying at least one or probably many entities includes relationships? Does it mean ontology is included in collection? Then what is explanation, super or upper ontology?     (2F15)

KenBaclawski: Ravi, a collection does not in itself include relationships. Relationships are specified with functions. An example of a subsumption relationship is shown in a later slide.     (2F16)

[12:17] DouglasMiles: Ken is getting into more the type theory foundation of logic huh?     (2F17)

KenBaclawski: Yes, this is type theory, but it is being used as the basis for explanations.     (2F18)

[12:17] ToddSchneider: John, Okay. Then how does the agent (receiving the answer) decide the received statement/assertion is an answer?     (2F19)

[12:18] Mark Underwood: Is "Explanation Logic" a semi-coherent discipline? I didn't run across that in the lit I was skimming     (2F20)

KenBaclawski: The term "Explanation Logic" was just a pedagogical device for this particular talk. It is actually a typed lambda calculus.     (2F21)

[12:19] DouglasMiles: All good that Ken is.. just untypical of this group     (2F22)

[12:19] John Sowa: Todd, the person who asks the question must always have the option of asking a follow-up question,     (2F23)

[12:20] John Sowa: The explanation is satisfactory when the person who ask the question says "Oh, I see" or something equivalent.     (2F24)

[12:23] RaviSharma: Ken_ man to mortal why? where is the why in this unless we understand from context or experience that all Men are mortals     (2F25)

KenBaclawski: The syllogism example shows only the explanation of why Socrates is mortal. The answer to why all men are mortal is a different question. It could be regarded as being a "followup" question. The explanation for why all men are mortal could be developed in much the same way, but ultimately there will be axioms (assumptions) that must be accepted by the parties if communication is to be possible at all.     (2F26)

[12:23] John Sowa: Mark, the questions that trigger an explanation are always metalevel questions.     (2F27)

[12:24] John Sowa: An ordinary Q/A system answers questions that can be answered with ordinary data.     (2F28)

[12:24] RaviSharma: Ken- What is the relation between collection and category in terms of metafunction?     (2F29)

KenBaclawski: Collection is part of the logic. Category is a metafunction about the logic.     (2F30)

[12:26] John Sowa: But an explanation system answers metalevel question, such as how, why, what is the evidence, who said that, why did they say it...     (2F31)

[12:26] RaviSharma: by explanation similar to my earlier Q all living things (beings) are mortal, man is an instance of that?     (2F32)

[12:26] Mark Underwood: John - I'm mulling that     (2F33)

[12:27] ToddSchneider: Jan Lucasiewicz's book "Aristotle's Syllogistic from the Standpoint of Modern Logic", Oxford Press, 1958, is an interesting read.     (2F34)

[12:27] Mark Underwood: RE Research threads in my talk address issues of missing, incorrect, dissimilar or highly domain-specific notions of who/what etc Socrates is and what is meant by asking if he is a man     (2F35)

[12:29] John Sowa: Mark, those are all metalevel questions.     (2F36)

[12:30] John Sowa: Short point: You can't ask a metalevel question in SQL, SPARQL, ...     (2F37)

[12:30] DouglasMiles: CYC having several proving modules that use different proof theories , afterwards it is important to create a proof that explains why each module was allowed to make such proof steps in the larger proof     (2F38)

[12:30] John Sowa: But it is possible to annotate data with metadata.     (2F39)

[12:31] John Sowa: In that case, you can use SQL or SPARQL to ask ordinary questions about the metadata.     (2F40)

[12:31] DouglasMiles: John: yet no one ever gets to that level in Sparql     (2F41)

[12:32] John Sowa: But then you would like to ask meta-metalevel questions about the reliability of the metadata.     (2F42)

[12:32] DouglasMiles: (but good to admit it _can_ )     (2F43)

[12:32] Mark Underwood: Actually, I wasn't able to cover RST well... Put that on the Known Deficiencies list     (2F44)

[12:34] DouglasMiles: Will followup questions be just expanding subproofs?     (2F45)

KenBaclawski: Not necessarily. Expanding subproofs are important examples of followup questions, but one could also ask meta-metalevel questions as John Sowa just mentioned. No doubt there may be still other kinds of followup questions.     (2F46)

[12:35] ToddSchneider: Mark, could you 'expand' the acronyms you've used: 'RE' research, 'RST', etc.?     (2F47)

[12:35] Mark Underwood: RST = Rhetorical Structure Theory https://en.wikipedia.org/wiki/Rhetorical_structure_theory     (2F48)

[12:36] Mark Underwood: RDA Research Data Alliance (I think Gary is a collaborator)     (2F49)

[12:38] DouglasMiles: Rhetorical relations in RST I think is the use of "Underspecification relations" ?     (2F50)

[12:38] DouglasMiles: "Underspecification relations" are like in-Underspecified     (2F51)

[12:40] DouglasMiles: Is example from CYC of Rhetorical Relation ?     (2F53)

[12:41] DouglasMiles:     (2F54)

[Mon](genlPreds authorOfProgramWithAIS by-Underspecified)     (2F55)

[Inf](genlPreds causes-EventEvent by-Underspecified)     (2F56)

[Mon](genlPreds createdBy by-Underspecified)     (2F57)

[Inf](genlPreds doneBy by-Underspecified)     (2F58)

[Def](genlPreds eventOccursNear by-Underspecified)     (2F59)

[Def](genlPreds filterer by-Underspecified)     (2F60)

[Inf](genlPreds near by-Underspecified)     (2F61)

[Mon](genlPreds objectConfiguredInChemicalBond by-Underspecified)     (2F62)

[Def](genlPreds selectiveBarrier by-Underspecified)     (2F63)

[Mon](genlPreds translationAlong by-Underspecified)     (2F64)

[Def](genlPreds traverses-Partial by-Underspecified)     (2F65)

[12:41] DouglasMiles: the "senses of "by") are hacked into a fake illogical predicate in this case but by-Underspecified still has a real world reason to exist in some sense all those predicates overlap     (2F66)

[12:43] DouglasMiles: Rhetorical Structure Theory seems to be based on believing this     (2F67)

[12:46] DouglasMiles: (near ?X ?Y) (createdBy ?X ?Y) do have some overlap     (2F68)

[12:46] DouglasMiles: just in logic they may not     (2F69)

[12:48] Mark Underwood: Donna, Ken has thrown down the gauntlet for a better coverage of RST (maybe for the Summit) - It's somewhat deprecated in the psych community as having weak evidence, but let's not be intimidated by that.     (2F70)

KenBaclawski: Indeed, RST does have weaknesses, but it seems to be a good place to start.     (2F71)

[12:50] Mark Underwood: The discussion of explanation logic feels akin to the attempt to create secure software through formal methods     (2F72)

KenBaclawski: Yes, but it is more than that. Mathematical proofs are supposedly the "gold standard" for all explanations (not just for secure software), yet mathematical proofs in the literature are only informal narratives that attempt to convince the reader. They are not as rigorous as one may be led to believe, and mistakes occur in published proofs.     (2F73)

[12:50] Mark Underwood: Not exactly an overwhelming success, but the reasons are complicated     (2F74)

[12:50] DouglasMiles: in CYC we decide during proof normality checking time that by-Underspecified may have been "authored by" instead of "near by"     (2F75)

[12:50] AndreaWesterinen: Sorry to have to leave early, but I have a call at 1pm.     (2F76)

[12:50] AndreaWesterinen: Interesting discussion.     (2F77)

[12:50] Mark Underwood: Be well, Andrea!     (2F78)

[12:53] Donna: one group we might reach out to: https://ieeexplore.ieee.org/document/4052915     (2F79)

[12:57] Mark Underwood: yes, I saw that one (on the cutting room floor)     (2F80)

[12:57] Mark Underwood: A. Nakasone and M. Ishizuka, "Storytelling Ontology Model Using RST," 2006 IEEE/WIC/ACM International Conference on Intelligent Agent Technology, Hong Kong, 2006, pp. 163-169. doi: 10.1109/IAT.2006.114     (2F81)

[12:57] Mark Underwood: Ontology -> agent-based tech     (2F82)

[12:58] DouglasMiles: +     (2F83)

[13:03] Gary: From Brian on the other/zoom chat 12:59 PM @ Ken, how well does your framework support resolution of meaning when dealing with multiple ontologies? -- e.g., Speaker applies ontology(A) and Audience interprets using Ontology(B) and (in the classical OR motif) G is unknown. I would also be interested in talking off-line about your work in ontology repositories and its ability to support my work on a multi-perspective calculus (spCalc).     (2F84)

KenBaclawski: That is a very interesting question, and I have done some work on this, but I could not resolve it adequately. The idea is to use theory morphisms to map one ontology to the other (if possible) or at least identify what the two ontologies have in common (using theory morphisms), and then build on this basis to resolve meaning as much as possible.     (2F85)

KenBaclawski: There is a form of OR that does not require a goal, and some logics support that approach, such as the Extended Calculus of Constructions.     (2F86)

[13:14] BobbinTeegarden: Peter Senge once said that what happens in a business meeting is only 20% what's on the table, the other 80% is what's in the bubbles over the participants heads. Contexts of contexts ... turtles all the way down.     (2F87)

[13:14] Donna: I need to drop off. Thank-you everyone. Ken - can you wrap up?     (2F88)

[13:19] MikeBennett: Mark you would be welcome to present this next week, from different hardware.     (2F89)

[13:19] Gary: An important part of discourse analysis relevant to explanation is identifying some of the higher level relations you find such as "provides evidence" or "refutes:.     (2F90)

[13:20] BruceBray: important topic but can only hear less than half due to audio issues     (2F91)

[13:22] Gary: It is usually the case that different background knowledge is being used by speaker and listener but exchanges try to converge these as "understanding."     (2F92)

[13:23] Mark Underwood: A use case for this is an oncologist "explaining" a diagnosis and treatment plan to a patient     (2F93)

[13:29] Mark Underwood: I'll talk about this diagram further     (2F94)

[13:29] Mark Underwood: Ack'd     (2F95)

[13:30] Mark Underwood: This is a third party sketch     (2F96)

[13:31] Mark Underwood: He's in a different ontology     (2F97)

[13:31] Mark Underwood: Be well, speak next week     (2F98)

[13:33] Gary: We need to have a better bridge explanation ourselves from the explanation domain that Ken talks about to what mark's domain of discussion of explanation is.     (2F99)

KenBaclawski: Agreed.     (2F100)

[13:37] BobbinTeegarden: What is the alternative to RDF right now? With tools support...?     (2F101)

[13:38] ToddSchneider: Alternative to OWL, RDF, ...: Common Logic. But its missing commercial support, for the moment. Industrial needs may change this situation.     (2F102)

[13:38] BobbinTeegarden: One can hope...     (2F103)

[13:38] ToddSchneider: Note that ontologies can be developed in natural language.     (2F104)

[13:40] BobbinTeegarden: @Todd ... and diagrams? A diagrammatic logic-al language?     (2F105)

Resources     (2G)

Audio Recording Due to technical difficulties, the audio for Mark Underwood's talk was difficult to hear so his presentation will be given again at a later session.     (2G1)

Previous Meetings     (2H)


Next Meetings     (2I)