From OntologPSMW

Jump to: navigation, search
[ ]
Session Universal Query Language
Duration 1 hour
Date/Time 11 August 2021 16:00 GMT
9:00am PDT/12:00pm EDT
5:00pm BST/6:00pm CEST
Convener KenBaclawski


Universal Query Language     (2)

Agenda     (2A)

Conference Call Information     (2B)

Attendees     (2C)

Discussion     (2D)

[12:28] Douglas R Miles: Questions for John: Are two quantifiers enough (donkey quantifiers?)     (2D1)

[12:40] MikeBennett: Surely much of the 'fuzziness' can be solved by distinguishing (a) 'Seat' - something you sit on, and (b) 'chair': something built for sitting on     (2D2)

[12:41] ToddSchneider: Mike, yes. To what level of detail is an entity distinguished.     (2D3)

[12:43] Jacob Friedman: including intension/extension     (2D4)

[12:45] Ram D. Sriram: @John Sowa: Do you have any references of the use of Zachman framework for ontology development.     (2D6)

[12:45] ToddSchneider: Ram,     (2D7)

[12:46] ToddSchneider: Ram, the usual interrogatives (i.e., who, what, where, when, ...) are themselves effective analysis tools for ontology development.     (2D8)

[12:48] BobbinTeegarden: John, how do you model context (w w w w w + h)? You showed Which (not part of the w+h string) and Why, where are the others?     (2D9)

[12:48] Douglas R Miles: speaking of 'fuzziness', it would be nice that people could give a value that says how far an instance is from the prototypical     (2D10)

[12:49] Douglas R Miles: (of course we have to define the prototypical)     (2D11)

[12:49] David Eddy: The implementer was a Marine.     (2D14)

[12:50] David Eddy: I was never able to get John Zachman to talk about that bottom, implementation row.     (2D15)

[12:51] RaviSharma: Hello chat is accessible to me now     (2D17)

[12:52] RaviSharma: David - the reason is that implemented enterprise is a living changing and organization practices based "living" thing?     (2D18)

[12:53] ToddSchneider: Bobbin, One approach to 'context' is use of the notion 'relevance'. What is 'relevant' to the 'immediate' entities or processes of interest (at a particular spatio-temporal extent)     (2D19)

[12:54] David Eddy: @Ravi... obviously formally capturing change is an essential ingredient for success     (2D20)

[12:57] David Eddy: Goes without saying Zachman "cell" framework is the language used in each cell tends to be somewhat unique. Something of a challenge to the base concept of "universal."     (2D21)

[12:59] BobbinTeegarden: @Todd, thanks interesting... but how do you physically model 'relevance' or context...?     (2D22)

[13:00] BobbinTeegarden: By the way, John Zachman was looking for help in turning his boxes into an ontology, any takers? Mike Bennett and I were talking to him about it a few years ago...     (2D23)

[13:01] Douglas R Miles: An important part of a "query language" is when a "Select person from table1 where bornAfter 1950" (loves ?J Mary) if the binding of both of these should be considered the same people or not     (2D24)

[13:02] Douglas R Miles: (FromDatabase "Joe") vs JoeTheGuy should be unifiable     (2D25)

[13:03] ToddSchneider: Bobbin, ahhh .... Details. 'Relevance' can be modeled as a numerical value between 0 and 1 (similar to a probability, if not explicitly a probability). But 'Relevance' also depends on the object/subject of focus (i.e. what is of interest) and another (potentially relevant) entity.     (2D26)

[13:06] RaviSharma: Zachman - framework which deals with all types of tools from requirements to systems that run enterprises as in defense, manufacturing, in and then I asked my Q to John     (2D27)

[13:06] RaviSharma: john will send references     (2D28)

[13:07] RaviSharma: john- each specialization will be thru dialog, tools in framework etc     (2D29)

[13:08] RaviSharma: same dialog at each level for 6 rows     (2D30)

[13:11] RaviSharma: john- jcl, cobol, etc legacy to do software re engineering Arun and Andre - Tailored to derive ontology and find relations? Accenture estimates were 80 human years     (2D31)

[13:12] RaviSharma: john was able to have these two persons do it in few weeks with ontogies, NLP, etc level.     (2D32)

[13:15] RaviSharma: developing a general dialogue system     (2D33)

[13:15] Andrew Dougherty: D3WA+ ?     (2D34)

[13:15] RaviSharma: simplifying to QA level     (2D35)

[13:16] RaviSharma: people in Enterprise can talk about computer or language     (2D36)

[13:16] RaviSharma: john's comments above     (2D37)

[13:17] BobbinTeegarden: Isn't Zachman more oriented towards the process, not the data (ontology) side of the business?     (2D38)

[13:17] BobbinTeegarden: You go ahead and ask...     (2D39)

[13:18] RaviSharma: john cited reference for Bobbin link.     (2D40)

[13:19] RaviSharma: a link in his references list.     (2D41)

[13:20] David Eddy: ... & humans will ask a silent question by raising an eyebrow...     (2D42)

[13:22] RaviSharma: Thanks Ken, we need to tie UQL with KG SQL?     (2D43)

Added later by KenBaclawski: @RaviSharma: KGSQL is specified in DOL, which ties KGSQL with UQL. KGSQL is also specified with denotational semantics which is useful for directly implementing KGSQL. I have a translation to SPARQL. which could also be used for implementing KGSQL. Incidentally, the reason why I asked for the CLIP specification in DOL is so that I could develop a translation to CLIP.     (2D44)

[13:22] RaviSharma: thanks     (2D45)

[13:22] BobbinTeegarden: John, are you building us a tool to do the interaction within a DOL framework?     (2D46)

[13:23] Douglas R Miles: Problem is these great Natural Language Understanding stuff is kept locked up     (2D47)

[13:23] janet singer: The topic Formal Ontology as Computer Artifact that Alex suggested would be a way to concretize issues raised here. The terms syntax, semantics, pragmatics are helpfully clarified by understanding that humans and machines both play roles in the artifaction context.     (2D48)

Resources     (2E)