Minutes of UBL Library Content Subcommittee

Tuesday, May 7, 2002, 8:00-9:52 am, California Time

Attendees

• Ron Schuldt
• Joseph Chiusano
• Tim McGrath
• Peter Yim
• Kurt Conrad
• Arofan Gregory
• Lisa Seaburg
• Marion Royal
• Sue Probert
• Debbie Quezadaz
• Mike Adcock
• Dave Lewis
• Sally Chan

Agenda

1. Welcome from Chair and appointment of Secretary to take minutes
2. Acceptance of previous minutes
3. Action arising from previous meeting
   • SALLY. Contact Frank and Arofan to provide the business process definition for SCM (also EAN/UCC/UMM).
   • TIM agreed to produce a next stage draft of the Process/Methodology paper and to send this to Lisa.
   • MIKE. Continue with the Dependencies paper.
   • ALL. Contribute thoughts about rules, dependencies etc. to Mike.
   • ALL. NDR Global Attributes issues. It was agreed that the group should study the relevant material in readiness for discussion on next week's agenda.
   • To chase up Arofan on the context paper for discussion next Tuesday conference call.
   • Ask Arofan for progress report on his areas (Context Assembly, Tools & Techniques) in time for next Tuesday’s conference call.
4. Current Review - Editorial Team report
   • Comments to date
   • Thursday meeting facility
5. Work plan
   • Enriching the metamodel - Dependencies, global attributes, uids, presentation attributes
   • Develop and document the UBL process/methodology
   • Extending the model - SCM next documents (PO Response)
   • CCTS comments
6. Reports on status of other UBL subcommittees
   • Tools and technique sc (Arofan)
   • Naming and design rules sc (anyone)
   • Context drivers sc (Sue)
   • Context assembly sc (Arofan)
7. Reports on status of other related projects
   • Core Component Technical Specs (Lisa)
   • Core Component Supplemental Documents (Sally) Next meeting is May 13th]
   • Ontology sub-group (Peter)
8. Other Business
9. Next meeting.

Minutes

Agenda Item #1: Welcome from Chair and appointment of Secretary to take minutes

• Kurt Conrad appointed to take minutes.

• It was noted that this is Ron's first attendance. He asked that Peter's report moved up, as he had to leave early.
  • Not much to report. Trying to get someone to sponsor Leo's trip to Minneapolis. Getting some help from Marion. Leo is trying to get some sponsorship from Mitre. Expect closure this week.
  • Arofan made point of convergence of threads
    • Examination of ontology engineering (science)
    • The work that mike's doing on controlled vocabulary
    • Identification of elements for PO Response
    • The context drivers that Arofan mentioned
  • The context drivers were discussed in Barcalona. No one came forward to help with this work, just Arofan and Sue. The task is to come up with value sets for defining different business contexts. This is to be a pure classification scheme. There's a draft in existence. It describes the conditions of business (not auto-generation of :tags)
  • Sounds similar to UDEF. Ron would like some time on the agenda to present a brief overview.
  • Tim see's convergence. Suggests an agenda item for the next face-to-face to pull these threads together. Between now and that meeting, he asks for some examples to be put together. Expects that it would be too hard to rationalize these topics remotely (i.e., outside of a face-to-face meeting).
  • The end goal is to not have different names for the same things across industries. Variations are to be handled via a contextualization mechanism.
  • The goal is not to invent new words, but to identify a stable set and articulate precise meanings for them. This group shouldn't be concerned with industry-specific naming.
  • Expected that UBL was not to be a 100% solution, but only the 80% core.
    • This depends on the definition of core.
    • "Core" is probably the mean (most commonly used).
    • Thought that agreed on 80/20 rule.
    • But want to avoid debates whether every element in library is used by all industries.
    • Something of an evolutionary process.
  • UDEF is available from web site. It's not complete, but represents a starting point. Objects for context and properties that tie to ebXML representation work. A URL is on the LCSC web page
  • Also some discussion about whether numeric representation of UDEF could work as unique IDs for elements in our library. Current thought is to keep this data as an additional column in our model instead of unique identifier.
  • Need good universal classification schemes for contextalization, but none really exist.

• Back to the proposed agenda

Agenda Item #2: Acceptance of previous minutes

• Change reference to Minneapolis
• Have actually completed 1/3 of elements

Agenda Item #3: Action arising from previous meeting

SALLY. Contact Frank and Arofan to provide the business process definition for SCM (also EAN/UCC/UMM).

• Have a list of docs:
  • Order
  • Order Response
  • Order Change
  • Event Shipping Notification
  • Proof of Delivery
  • Invoice (for both procurement and supply chain)
• VMI stands for Vendor Managed Inventory
• CPFR is Collaborative Planning Forecasting and Replenishment. Would include forecast and inventory report.
• For business process, we have the EAN example. We can reference the EAN example for the business process. Frank said that. Sue suggested looking into EAN example.
• SALLY: Could you prepare a 1 page doc that summaries what you just said? Would it be unreasonable to ask that this be presented as a business process model? Need to break that stuff down, as part of the context work. Arofan can do. The biggest context driver is business process.
• Arofan: Sally, send stuff. I'll (Arofan) put together a summary and will verify appropriate level of detail based on requirements.
• Sally will send it to Arofan

TIM agreed to produce a next stage draft of the Process/Methodology paper and to send this to Lisa.

• Tim has done something, not exactly the task: Recirculated the existing doc to Sally and Mike to support next revision
• One of outstanding issues is the chestnut issue of code vs IDs. Still need to have a position on it.
• While haven't done exact task, have kept debate alive. No next stage draft until we resolve these issues.
• Was the purpose of sending to Lisa so that she could do some work on it? Document started with Lisa
• Document isn't accurate, but not sure what it should say

MIKE. Continue with the Dependencies paper.

ALL. Contribute thoughts about rules, dependencies etc. to Mike.

• Have you had any feedback?
• Have done no further work. Haven't seen any further email input to dependencies paper.
• This topic came up as part of the comment against the core component technical specification. It includes some of concepts that we are talking about (description of Core Component types and supplementary components). There's some question about whether these things are within scope of paper.
• Naming and design rules people also looking at these issues (relationships between core component types). There's been some discussion.
• Hope that Billy will pass his ideas on to you. Most of this is Mike running with this on his own. Bill has voiced his opinion, but it is not really in the NDR area and not to be discussed there (other than how to represent the metamodel after it is stabilized).
• A good source to summarize? Get comments that went to CCC group. Bill is the man. Tim will forward Bill's inquiry.

ALL. NDR Global Attributes issues. It was agreed that the group should study the relevant material in readiness for discussion on next week's agenda.

• This was left on by mistake. Actual issue was called Global Attributes. Really common attributes (i.e., additional columns in our model). There's a difference between the logical model in the spreadsheet and the physical model in the schema. Involves decisions about how to represent things in schema syntax. Those rules are currently hard-coded in perl scripts. No one thinks that it's a good idea. The proposal is to capture this info as part of spreadsheet. Gunther has produced a paper that described metadata that he thinks that he needs.
• Is Gunther talking facets?. No, covers the uid/uidref/xlang type of stuff.
• There's another issue being discussed by Gunther and Arofan. Arofan will raise it later in the telecon. When we get to work plan, we can talk about Gunther's paper.

To chase up Arofan on the context paper for discussion next Tuesday conference call.

Ask Arofan for progress report on his areas (Context Assembly, Tools & Techniques) in time for next Tuesday's conference call.

• Is there a draft of context paper?. Yes, but it is very rough. Sue and Mike had been working on harmonized list of role codes. Sue had collected some info. That is a total gap in the paper. What Sue had collected wasn't very helpful..
• There is a draft for people to look at. Arofan needs to send it out, but it's not ready for the web page.
• The paper focuses on classification. The first few pages might be helpful, but the rest is reproductions of various taxonomies. The only fun thing that can do with thee taxonomy is to see if you can reasonably describe your business with the values provided.
• Tim wants to see it to get an idea of how the context methodology group is thinking.
• The document is rough, just a pile of initial stuff. Lisa will include in working group subdirectory with appropriate disclaimers.

Agenda Item #4: Current Review - Editorial Team report

Comments to date

• Closing date for comments is next Thursday
• Lisa has sent acknowledgments to those who commented and compiled the comments into a single document w/ the author's name attached to each comment. The next step is to go through comments (Lisa, Peter, Sally) and get them ready for the whole group to go through them
• Have received 5 sets of comments. The Liaison Committee should put some work into getting more comments
• Arofan was to send in a comment for xCBL: "No definition of data types. Want to know specific numeric forms to do such things as price reconciliation." Arofan to write up and send in. mention issue of precision. Tim's aware that the issues are being discussed, but needs a formal submission put on the table for tracking
• In terms of Liaison Committee, Tim has telecon with people that tried to talk to in Barcalona (AICX?). Expect them to give a set of comments/questions, but unlikely to get input in by 10th. May want to consider accepting comments after cut-off
• Since the LCSC hasn't really set up verticals, does it make sense to come back to them at a later time. Yes, but there's an additional benefit of asking for comments at early stage to introduce them to our thinking (to validate not just the content, but also the process) and have them begin to validate reusable components that are of interest to them
• There's another LCSC/Liaison Committee meeting this Friday to make a last push for comments.
• Tim doesn't have a problem with people providing comments at any time. It's just a question of what we are prepared to do with them at any given time (scheduling).
• A major item at the face-to-face is to decide what to do with comments so that we can give feedback
• Anything else for Sue? She sent out ATA Purchase Order. Tim also shows Sue, Mike and Tim working Order Response documents
• Registration required by X12 (05.24), also a closing date of 05.10 for hotel discount

[Sally departs]

Thursday meeting facility

• Editorial Review Team: Any response to Thursday meeting? Tim sent email to Polly. If want Thursday meeting, Tim should set it up. Tim will schedule it for 9:00am California time

Agenda Item #5: Work plan

Enriching the metamodel - Dependencies, global attributes, uids, presentation attributes

• Already had a bit of a discussion on Mike's dependencies and global attributes (which is a non starter)

• Haven't discussed uid. Gunther's position paper. Having read Joe's interpretation, perhaps Joe should summarize
• Each of Core Component entities would have 4 attributes (uid, uidref, uidrefs, xml:lang). Don't believe that uid is only an ID type. It has to be unique more globally. Also have to relate Core Component back to BIE. Believe that Gunther intends to make associations using uid and uidref.
• That is one of several applications. uids also function as plain old uids for any referencing. We probably saw same problems in uid
• Schema has key/keyref that overcomes a few flaws. 1) ID values have to be unique. 2) Processor can't verify that an association actually exists (only that IDREF exists). 3) ID attribute can't begin w/ a number. Key/keyref overcomes these problems and can values can be bounded using regular expression. Can also bind references to specific element types
• If give IDs to all constructs in library, what if something comes along that needs to be unique but is an extension to UBL?
• Discussion indicates that have identified a requirement to maintain the referential integrity of our model and the mechanism to do that is not the ID/IDREF mechanism, but XSchema key mechanism
• Discussion also identified that need to have a more detailed granularity of the types that we use. The current issue of having a core component type (now just type) is inadequate for our requirements. Need to identify the content components and supplementary components. There's a need to maintain referential integrity, but don't capture that info
• Need to establish the types of IDs and references. Already given types names and defined things using types, don't need referencing for that. Need it for relationships on the business side
• The registry will also help out, given associations in the registry
• There's a question of implementation and question of content. Just want to be comfortable that we're capturing the content that we need. Concerned that there's a lower level of detail that we are now talking about. Want to see a worst example of how it would look. See a great danger in maintaining both ID/key references and name references
• There will be cases where we don't have a name reference and need a fallback (like things in registry model that we currently can't support). We're currently in good shape with regards to registry
• Do we need to worry about this? Not yet. Haven't defined a specific problem that needs to be solved

Develop and document the UBL process/methodology

Extending the model - SCM next documents (PO Response)

• With respect to extending the metamodel: Started to identify structures within Order Response documents that are new and modifications of what already have.
• Is modeling based on the process model that Sally has specified? No, don't really have a process. Starting w/ xCBL Order Response, adding or removing structures as a process model dictates
• in CBL, the order response is used in response to Orders and Change Orders. There's a lot of change order garbage that could/should be dropped.
• Yes, we're not interested in building umbrella documents. May want to make that statement very directly, it's different from what some folks may expect

CCTS comments

• Hoping for a lot of "please explain" comments from the editing team. Very pleased with what submitted, even if not all of it is accepted. Made more sense after being rewritten

Agenda Item #6: Reports on status of other UBL subcommittees

Tools and technique sc (Arofan)

• Tools and technologies mostly Gunther's maintenance. Haven't seen much of him. He's on vacation a lot. Also his email address has been taken off of the list server today.

Naming and design rules sc (anyone)

• Who attended the call last Wednesday?. Eve said that we would discuss the proposal one last time. The committee defined its fallback position / contingency plan. That and discussion about context extension.
• In terms of the face-to-face, we should schedule some joint meetings with NDR. Tim will approach eve to discuss that
• namespace versioning: becomes relevant w/ regard to order response

Context drivers sc (Sue)

• Outstanding issue is the metamodel / transformation that deals with facets. If we need to effectively capture additional info, now it a good time to specify the requirements
• Its a lot easier to work to an existing spec. There's this idea of facets, where people can specify the date/time format that they use. We need to make sure that there's need for these mechanisms, not just "cool"

Context assembly sc (Arofan)

• Nothing to report
• Context assembly having regular meetings, getting very deep, but don't expect answers until after face-to-face that will impact LCSC.

Agenda Item #7: Reports on status of other related projects

Core Component Technical Specs (Lisa)

• Some LCSC members will be working at Barcelona EWG meeting. Haven't seen a schedule yet. Lisa's not going to Barcelona. Arofan is willing to fill in. Just help Mike if he needs help
• Also expect some feedback from the editing group

Core Component Supplemental Documents (Sally)

• Haven't met since last week.
• Peter talked about the ontology subgroup and hope to get Leo over for the face-to-face
• The next meeting is May 13th

Ontology sub-group (Peter)

Agenda Item #8: Other Business

• Mark has asked for resource requirements. Need to order the right number of projectors and power sockets. Now putting down that will have 20 attend. Expecting reasonable turnout
• Any phone conferencing capabilities? Leo could give a tutorial remotely

Facets Issue:

• NDR had a discussion about different string lengths, numeric formats, date/time, etc. Gunther suggested providing a mechanism for you to describe what kind of data format is being used (basic datatype stuff). Will either have to do a conversion or not understand the info.
• The danger is that if give freedom to specify, it cuts into interoperability.
• Believe that have to give ability to specify, but have an extension mechanism to change what's in the core library. Argument is to provide defaults for people to change. Don't have to use extensions out of the box
• For core component types, may have to do some subtyping. Would also drive request for LCSC to define default data formats for this stuff
• Understand issue, but need to not view as an isolated requirement. Ties back to business rules. Also expect that most common scenario isn't to maintain their internal formats, but things like bank ID, where there are various standards.
• There's an issue of 3-way matching. process being limited by the lowest level of precision. Forcing 3 digits w/ precision of 2 doesn't help. At least have a starting point. Whether you call them defaults or core definitions, there are potentially several for each type.
• Still intrigued by Core Components vs supplementary components. Those are very related concepts. Care that we get some sort of resolution / recommend some basic approach for doing this
• It may be that Core Components don't take to sufficient level of syntax specificity
• The proposal is to not just have number, but price type. To come up with all of the subtypes necessary for a business document
• The issue is at what layer do you draw a line (defend universally vs defined by us)? Expect that they will define what they define and we are probably the only group to define what we need. Can provide feedback to CC committee
• Asking whether library subcommittee ready to define defaults that want to see, regardless of how to express those defaults
• See overlap in work that Eve is looking at. Need to tackle together. Know that there's something out there, but don't currently have enough focus
• Can have different solutions to different things. Can make recommendations about dates and periods.
• In danger of solving one problem in several different ways. Might have to solve in different ways for different circumstances. Not yet convinced that it is one problem. Could be that what's been identified is part of a bigger problem
• Have to remove ambiguity, and need level of precision required to remove ambiguity. Need to push forward on discussion of dependencies. Objective is clear understanding of problem and what info we need to define to solve it. Then it becomes an NDR discussion about how to solve.
• That drives a parallel issue, we can make it really hard on XML designers. If we expect the parser to validate data, it can create implementation issues
• One goal has been an unambiguous logical model that is syntax neutral. [is syntax neutrality limited to tag structure, or does it include low-level data encoding?]
• Need to get this on agenda for the face-to-face for all people who have an interest in this. Want one work product to be the table model that gives us what we need for the next generation
• Want to keep discussion going prior to the face-to-face, perhaps a position paper or someone who can talk to this issue
• Mike and Arofan own this issue for LCSC and NDR

Agenda Item #9: Next meeting.

• This has been the best attended meeting for a while
• Next call is at this time on 14 May 2002.