Actions

Ontolog Forum

Session Planning
Duration 1.5 hour
Date/Time Nov 15 2017 16:00 GMT
9:00am PDT/12:00pm EDT
5:00pm GMT/6:00pm CET
Convener MatthewWest

Ontology Summit 2018 Final Planning Session

Agenda

1. Revise the summit abstract. The current version is at OntologySummit2018#Description

2. Matthew West will be convening a session on contexts for integration. Slides Video Recording

Conference Call Information

  • Date: Wednesday, 15-November-2017
  • Start Time: 9:00am PDT / 12:00pm EDT / 6:00pm CET / 5:00pm GMT / 1700 UTC
  • Expected Call Duration: ~1.5 hours
  • Video Conference URL: https://bluejeans.com/703588230
    • If you have not used BlueJeans before, then connect to the URL above before the meeting time so that the required plug-in can be installed.
  • Chatroom: http://webconf.soaphub.org/conf/room/ontology_summit_2018
    • 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.
  • This session, like all other Ontolog events, is open to the public. Information relating to this session is shared on this wiki page.
  • Please note that this session may be recorded, and if so, the audio archive is expected to be made available as open content, along with the proceedings of the call to our community membership and the public at-large under our prevailing open IPR policy.

Attendees

Proceedings

[9:43] David Whitten: Hello Ken. I'm looking forward to Matthew West's talk. He is a great advocate of representing time in a four-dimensional way.

[12:02] David Whitten: Matt is getting ready to present his talk. "Context in Integration"

[12:04] David Whitten: The word "Harmonization" was disputed by Matt West. He feels "context" is the issue why Harmonization isn't possible. Each Ontology has context that forces the Harmonization issues. Dealing with them forces the wrong issues to be addressed.

[12:05] David Whitten: Matt feels Integration is all about Harmonization. Janet agrees that "Context is the reason harmonization is necessary"

[12:06] David Whitten: Matt feels you can only Harmonize within a particular framework. The 3-D time model and the 4-D time model are incompatible. There isn't a single harmonization that generalizes both 3-D and 4-D.

[12:07] David Whitten: Janet feels Jeff was focused on diverse semantics between people, all four of the Human and Machine paths still need to be dealt with.

[12:07] David Whitten: Matt feels that the extended subtitle was less understandable than the base title proposed.

[12:09] David Whitten: Janet likes "Contexts in Context for Ontology"

[12:09] David Whitten: Matt likes simply "Contexts in Context" - Ken notes that Pat Hayes already took that title. (Matt thinks it is a clever wordplay)

[12:10] David Whitten: Matt feels that if you don't have requirements then you have no point of making an ontology to begin with.

[12:11] David Whitten: I think this points to the "intent" issues that were brought up by John Sowa.

[12:11] John Sowa: I like "Contexts in Context"

[12:12] David Whitten: I'm not even sure what title we are talking about. The Ontology Summit title? The title for the talks that Matt will give?

[12:13] David Whitten: Matt thinks this would be a subtle compliment to Pat Hayes' paper.

[12:13] KenBaclawski: http://ontologforum.org/index.php/OntologySummit2018#Description

[12:13] David Whitten: It has been clarified we are talking about the Ontology Summit Title.

[12:13] David Eddy: I'll vote for "Contexts in Context"

[12:14] David Whitten: John Sowa says the title: "Ontology Summit: Contexts in Context" obviates the need for the "for Ontology"

[12:15] David Whitten: Matthew West is talking (this evening) about "Context in Integration"

[12:16] David Whitten: He will talk about Integration and how models are used for systems and organizations. This is an easy confusion that will soon have folks talking at cross-purposes.

[12:17] David Whitten: afl

[12:17] David Whitten: afk

[12:19] David Whitten: back

[12:20] David Whitten: Off-line messaging requires that the common model is in the messaging system. Each of the communicating systems 1,2,3 make their internal model match the messaging model. (btw this is HL7's model)

[12:21] David Whitten: In the off-line messaging model with a facade, you have a separate system that uses the messaging models as a form of interface for each of the main models.

[12:22] David Whitten: The Shared database keeps a copy of the 1,2,3 with the messaging system internal to the shared database.

[12:23] David Whitten: John Sowa points out that the "shared blackboard" would externalize the messages, and would involve a system putting a message on the blackboard, which is reviewed and answered by any system which has access to the blackboard.

[12:24] David Whitten: The "Model Integration" slide has a common Integration Model database with private model data-models for each system that uses it.

[12:24] David Whitten: Matt agrees that John's model is good, but points out the slides he has are only a beginning point for the huge number of real models

[12:25] David Whitten: Matt states that the Integration model is always available. With more than one integration model you can isolate models of some parts of the system.

[12:26] David Whitten: You can even have a hierarchy of integration models which on the surface looks good, but may really complicate the process of communication.

[12:27] David Whitten: The "limited integration model" means you leave out all the context markers for the two application models that share a context (slide 13)

[12:28] David Whitten: The scope of a model is what entities the model covers and the Context is an idealized boundary which includes those entities.

[12:28] David Whitten: Integration Models are inherently brittle. A new definition of context forces re-creation of the Integration Model.

[12:29] janet singer: Context as what was left out - similar but different in human-human, human-machine, machine-machine communication?

[12:29] David Whitten: slide 15: as soon as you say "this is so broad a context we will never need to increase its size" you will get an example that breaks it.

[12:30] David Whitten: If you treat your integration models as sole support of the particular non-explicit context you can treat them as puzzle pieces (slide 16)

[12:30] David Whitten: You must make contexts explicit and then things work better. (slide 17)

[12:32] David Whitten: slide 18, a Federation model where everyone does whatever they wish is very brittle because a change in one system requires all the other systems need to change their interface. This situation becomes quickly un-maintainable.

[12:33] David Whitten: Slide 18, The fix for Federation models might be seen as a "Virtual Integration Data Bus" which is difficult to deploy, and which requires a common language.

[12:34] David Whitten: Slide 18: the Federation model only allows one location (each of the systems) for computation. If you move all the models into a monolithic integration model, you will have multiple locations that can process and have a common external model. This does put all the eggs in one basket.

[12:36] David Whitten: Jack Ring points out that this model involves a common time clock or time basis. Matt says he is only pointing out the classes of integration and the time issues. Jack Ring feels many software bugs can occur from this. Matt feels the push-pull dichotomy of interfaces can deal with some of these times issue.

[12:38] John Sowa: Integration is important for a single enterprise. But it's important to support interoperability among heterogeneous systems from many different enterprises.

[12:38] David Whitten: Matt says that the integrated ontology format is an interface for every system and issues of data conversion is a function of each system's interface. This should be a secure interface that makes the external world acceptable to the internal world.

[12:38] janet singer: @Jack - The overall title is now going to be simply "Ontology Summit: Contexts in Context" so we should reclaim your subhead for our track.

[12:38] John Sowa: One very important example: Anybody who wants to sell anything via Amazon must conform to their database schema.

[12:39] John Sowa: The Amazon schema has become the international default for selling anything.

[12:39] John Sowa: Where does the Amazon schema fit into this framework?

[12:40] David Whitten: Matt points out that many people use monolithic integration. The limits of monolithic integration are that making explicit issues of context that your application doesn't care about is expensive.

[12:42] janet singer: @Jack - Since we are under the heading 'Domain specific needs for context' our track could be 'The challenge of harmonizing diverse conceptualizations in systems engineering and realization'

[12:42] janet singer: @Jack - What do you think?

[12:46] David Whitten: "Primitive Concepts" foundation concepts are very rarely used by folks writing an application. These general concepts might be those used in "Physics models" Specialized forms are very common, equivalent to a screwdriver or hammer. A Cathode Ray tube or centrifugal pump are very discipline specific.

[12:47] David Whitten: Slide 26

[12:47] David Whitten: Consolidation is much easier when the internal identifications are mapped to external ontology models.

[12:47] David Whitten: This is hard work, but necessary. There is no silver bullet.

[12:48] David Whitten: Domain Ontologies take a long time even when they are dependent on top level ontologies.

[12:48] David Whitten: A top level Ontology (TLO) must be designed to allow domain ontologies to do specialization.

[12:49] David Whitten: John Sowa points out that the Amazon "ontology" has no upper level ontology.

[12:50] David Whitten: I note a list of words without an upper ontology is basically a vocabulary not an ontology, as there are no groups.

[12:51] David Whitten: Matt thinks that Amazon has a model that isn't good at anything but sales. The limited application model is something that can be expanded greatly. They are putting off specifications, and forces the customers to create the higher level ontology.

[12:52] David Whitten: Matt points out that slice 21 has some of these issues.

[12:52] David Whitten: John Sowa feels that you can have interoperability between many systems, but it isn't the integration issues.

[12:53] David Whitten: Matt feels Amazon only provides a limited set of data that only they decide is important.

[12:53] David Whitten: They don't care about anyone else's organization. (they have a derived concept set such as the "red" side of the triangle.

[12:54] janet singer: @Ken - Can you rename our track 'The challenge of harmonizing diverse conceptualizations in systems engineering and realization'

[12:54] David Whitten: Matt says Amazon expects someone else to deal with this issue as they treat this as an "someone else's problem"

[12:55] David Whitten: Matt and John point out the upper ontologies are requirements for integration.

[12:55] David Whitten: John people don't inter-operate on their high levels.

[12:56] David Whitten: John: each location has their own model, Walmart, sears, etc.

[12:57] David Whitten: John: They are most interested in minimizing their cost. They only have an agreement at the linguistic level of "words". With vague and flexible agreements, the brittleness goes away.

[12:58] David Whitten: John thinks you can interoperate at the application level without the applications being integrated.

[12:59] David Whitten: James Davenport says: Yea, verily, interop and integration differ in that integrators are striving to create a new system whereas interopers are striving to produce an effect.

[13:00] KenBaclawski: @janet singer: Okay, I changed your session. Unfortunately, I have to leave right now. If any of the tracks would like to schedule a planning session, please let me know, and we can set up a meeting for you after the US Thanksgiving Holiday.

[13:00] David Whitten: John says we must recognize heterogeneity is the norm. He agrees with John Lenat (Cyc) and Barry Smith (BFO) that the issues of top level ontologies are less useful. The top level is useful for design.

[13:01] David Whitten: Jack Ring has a bad echo when he talks.

[13:02] AlexShkotin: Matthew has echo

[13:02] David Whitten: Jack Ring considers the lack of interoperability/integration is a technical debt for software.

[13:03] AlexShkotin: switch his micro, Ken.

[13:03] David Whitten: John says interoperability doesn't require a common top level just that the specific middle level needs to be equalized.

[13:03] David Whitten: John feels the lowest level standards are at the ground level of your ontology.

[13:03] DouglasRMiles really likes Davids notes

[13:03] David Eddy: excellent JS & MW discussion

[13:03] David Eddy: & JR

[13:04] David Whitten: Whether you use English measurements or Metric measurements this is only has limited applicability.

[13:04] AlexShkotin: @Matthew switch off your microphone please

[13:04] David Whitten: John feels Mathematics is universally common, but doesn't say anything about what really exists.

[13:05] David Whitten: John thinks the major geometry used today is Euclidean. It specifies very specific system that can be mapped to Cartesian system or some other.

[13:06] David Whitten: John feels that communication conventions must be separated from ontologies.

[13:06] ToddSchneider: Have to go. Cheers.

[13:06] David Whitten: Janet says you don't need semantics if your pragmatics is in common.

[13:06] AlexShkotin: @Ken could you please switch off Matthew West

[13:07] MatthewWest: My phone line went down, but strangely I still have broadband.

[13:07] DouglasRMiles: It really is an echo .. it encode John, Janet, and Jacks voice very differently

[13:07] David Eddy: I see MW now muted

[13:07] AlexShkotin: Thanks

[13:07] David Whitten: Janet wonders if Amazon is dealing with Semantics. Are they just defining a particular syntax ?

[13:07] David Whitten: John feels Amazon uses a semantics based on WordNet.

[13:08] David Whitten: John says word net's grouping is "good enough" for many communication systems.

[13:08] MatthewWest: OK. The reason a TLO is useful in interoperability is that it means that you can simplify the interfaces to a common messaging system ( see slide 1.

[13:08] MatthewWest: Slide 17

[13:09] David Eddy: WordNet doesn't even include JCL

[13:10] David Whitten: John feels the integration of heterogeneous data will requires someone to spend lots of money which might kill the ontology fields work in formal logic.

[13:10] David Whitten: Matt, did you stop sharing?

[13:11] David Whitten: Janet says a flexible agreement is necessary across multiple enterprises and stakeholders

[13:11] David Whitten: John reminds us that C.S. Pierce it is easy to get agreement if you are sufficiently vague.

[13:11] MatthewWest: John, you are not looking at what I am saying. Look at Slide 17. You are assuming monolithic integration is the only form of integration. With a messaging system approach, the systems you integrate are unaffected (do not need to adopt any TLO) but interoperate using a messaging system that does use a TLO. (You can have multiple of these of course).

[13:12] David Eddy: In context of systems interoperability... very, very few systems owners have a firm grip on what their terminology / concepts / terms / acronyms / etc. actually mean... PRECISELY

[13:12] MatthewWest: David, I have to stop sharing when I use Soap Hub.

[13:12] David Whitten: John feels that very precise formal independent systems can only communicate with vague common sense notions of the terms.

[13:13] David Whitten: John advocates defining a method for systems to ask questions that will clarify definitions.

[13:14] David Whitten: Jack feels it is possible to inspect the heterogeneous for the presumptions of the programmer for software systems.

[13:15] David Whitten: Douglas Miles things that the issue of extracting these assumptions is where we want to get to.

[13:16] David Whitten: John says there is a top-500 company that has 40 years of documentation, software, reports, e-mail, IBM JCL, SQL, COBOL. they analyzed and cross-referenced all the common reports. Analyzing all discrepancies they could find. Glossaries for humans, etc.

[13:16] BobbinTeegarden: Isn't that what ETL in effect is doing? But with straight bottom level mapping, no ontology...

[13:17] David Whitten: John said it is a huge mess. It took 2 weeks of computing. But it means that even when the precise definitions existed, everyone used the common definitions.

[13:17] David Whitten: Matt agrees with John.

[13:18] David Whitten: @Bobbie, I don't know what ETL is. (or even what its acronym stands for)

[13:21] David Whitten: Matt points out on slide "Federation and Integration" that there is not requirement that there be only one Data Bus, but that that each Data Bus must have a common communication system.

[13:22] David Whitten: Matt feels that "Amazon" is a blue box on that slide. Each system that communicates with Amazon decides which model to use.

[13:23] David Whitten: The accountant system has its own model. Its not about taking over the world. It is about making a system that communicates with the world.

[13:23] BobbinTeegarden: ETL (Extract, Transform and Load) is a process in data warehousing responsible for pulling data out of the source systems and placing it into a data warehouse. It's also used to harmonize disparate (legacy and emerging) systems.

[13:23] David Whitten: John feels when you are looking at the Internet as a whole, only the TCP/IP is common. The TCP/IP is pure syntax.

[13:24] David Whitten: John points out the IRS model is one that all business and individual must conform.

[13:25] David Whitten: Matt says IRS is a de jure standard. The Amazon is a de facto standard.

[13:25] David Whitten: Matt says these provide an application model.

[13:26] David Whitten: Doug Miles think that people use whatever model they need to accomplish things.

[13:27] David Whitten: Doug says using the Matt and John are talking about different things.

[13:27] David Whitten: Jack Park wants a session of Integration vs Interoperability.

[13:28] David Whitten: Jack Ring (not Park) wants Matt and John to create this session.

[13:28] Gary Berg-Cross: have to leave....

[13:29] David Whitten: John says that Word-Net is the most common semantics, and is only used for interoperability, not for integration.

[13:30] David Whitten: Matt feels he is talking about multiple Top-Level Ontologies.

[13:31] David Whitten: John feels the modules you use to create the Ontologies is more important.

[13:32] David Whitten: They both agree that Whole-Part and Time (3D+1, 3D, & 4D) are important that they be explicit.

[13:33] David Whitten: Matt says we are doing the work of the summit. ; shared service, interoperability, and integration are important.

[13:33] David Whitten: these are issues of cross-purposes

[13:34] David Whitten: Janet points out that the issue of people in the picture, for machine-to-machine it must be exact. The broad picture that includes people and machines should be a hybrid system.

[13:35] David Whitten: Jack Ring points out that humans who understand each other are different from humans that act like no one else exists.

[13:36] David Whitten: John says that a precise data definition is useless when people who read output and create input using Natural Language definitions.

[13:37] David Whitten: John Sowa thinks that scientists, and specialized medical people are using specialized dialects (jargons)

[13:38] David Whitten: Doug Miles feels that the CycL feels that its formal logic is very readable. (John says using Phrases as single units with a formal definition is something Cyc has done that is very useful)

[13:39] David Whitten: Doug Miles suggests that even when people create Cyc Terms that are specialized for a different context, but there is a common meeting that collapses terms. Doug says CycL as a richer superset of English.

Resources

Video Recording

Previous Meetings

... further results

Next Meetings

... further results