ATTENDEES: Wendy, Jon, Dan S., Jeremy, Larry, Flavio, Arofan
Summary document: This is just a summary of the discussion that took place, not a fill-in all the details of what each product is etc. So really a general road map. 1 - is it accurate 2 - is anything lacking 3 - sending to Exec
Action item for DDI Alliance: Name of DDI4-Core needs to be determined in terms of the product line
SDTL support: 3 points (see TC-212) 1 - do we support this as a DDI product 2 - positioning within the product suite 3 - outline for setting up the group within DDI (what's needed...id persons, linking into github DDI site, oversight role of TC, etc....this may change with reorganization of ScientificBoard)
DDILIFE-3679 replace the ability to reference complience statements that was accidentally removed Jeremy will create pull for that in the next couple days
ATTENDEES: Wendy, Jon, Dan S., Flavio, Larry, Jeremy
Updated on EDDI event Reviewed outstanding issues and resolved or closed specific issues related to schemas The remaining issues have to do with pulling together and getting out 3.3 for final review Goal is to have everything together for a final approval for review release on Jan 9. Meeting on Dec 19 dependent upon having any agenda items No meetings December 26 or Jan 2
ATTENDEES: Wendy, Jon, Oliver, Larry
Reviewed all the remaining schema fixes and closed All remaining changes are to documentation Review of inline documentation will be made tomorrow
The documentation will be stripped back from user guide taking out element specifics which will be available in inline DocFlex or content generated by sphinx
Schema works looks done. Jon will prepare draft of announcement regarding Public Review and availability of Schemas Goal is to have this out the week before EDDI Public review to start as soon after EDDI as the high level documentation is in shape and continue through the end of January 2020 (minimum of 1 month but holidays mean extending that)
Issue filter for public - Issues addressed in preparation of DDI Lifecycle 3.3 Issues were reviewed during development of DDI Lifecycle 3.3 They provide information on the content, discussion, and disposition of the issue https://ddi-alliance.atlassian.net/issues/?filter=11223
Next week: Draft of announcement to DDI-Users list regarding Public Review and availability of schema
ATTENDEES: Wendy, Jon, Dan S., Jeremy, Larry, Oliver, Flavio
3667 some question in red within spreadsheet, draft of a change sheet to accompany if deemed useful Please review and comment on any changes 3669 resolved 3637 reopened with some examples - renamed to clarify content and moved to FUTURE (3.4) 3638 in progress 3636 resolved 3660 still to enter 3635 resolved 3670 resolved 3651 reopended and resolved
Verification to make sure no change was overwritten is down to 2 long lists of changes and the new SDI content to both verify and note where diagrams should be changed.
Oliver updated on Controlled Vocabularies: CESSDA now produces HTML with the language selection option as well as a show all languages. This in on internal at the momement but will be available. This means all we need is a tool to grab the three outputs for a controlled vocabulary, run the CodeList structure, zip and deliver
GOAL: complete entry of all resolved items and corrections found in verification work for next weeks meeting NO MEETING: November 28 (US Thanksgiving) and December 5 (EDDI)
ATTENDEES: Wendy, Jon, Jeremy, Dan S., Oliver, Larry
Comment on summary document Let Wendy know if there are issues (for those who haven't read yet) Post and send to Jared when confirmed TC-201 CV update updated others - we have an agreed solution with CV and ICPSR Oliver has talked to CESSDA about language bar on HTML DDILIFE-3664 in review closed DDILIFE-3660 information classification statement (if in review) not entered yet DDILIFE-3663 Best Practice Doc update (if in review) on hold DDILIFE-3665 package content agreement Jon has put on GITHUB ddilifecycle documentation - 3.2 and 3.3 together DDILIFE-3662 readme.txt specify changes changes reviewed in progress assigned to Jeremy DDILIFE-3667 change log - specify content in progress - assigned to Wendy Options listed DDILIFE-3659 developers documentation added notes - post 3.3 release DDILIFE-3639 create branch for 3.2 added notes post 3.3 release
added new label 3.3_publication
20192021 - 20191025
ATTENDEES: Attendees: Wendy Thomas, Jon Johnson, Larry Hoyle, Jeremy Iverson, Dan Smith, Johan Fihn Marberg, Oliver Hopt, Dan Gillman (as available), Barry Radler (Monday-Tuesday), Flavio Rizzolo (remote, as available), Jay Greenfield (remote, as available)
Small meeting today so we just checked for any outstanding questions regarding the October meeting. I linked in the following page with the start of the meeting agenda. This may move due to technical issues but its here for now.
Next week will be 3.3 entry for review and approval
ATTENDEES: Wendy, Dan S., Oliver
Small group. No major agenda items. Discussed organization and prep for COGS work and Oliver and Dan S. will look at to see if there is any change in the organization of activities needed.
Updated on meeting facilities.
Next week: 3.3 pull content review
ATTENDEES: Wendy, Oliver, Johan, Jeremy, Barry, Dan S., Flavio
The content of the meeting and organization was discussed in particular focusing on what needed to be done prior to the meeting in terms of individual preparation; content of the first day discussion and what the outputs of that discussion needed to be in order to inform DDI members and to specify activities for the rest of the week.
TO DO: wlt do asap
find and summarize documents on design approaches over time (to inform overall production approach discussion)
post Roadmap original and updates for individual review
prepare a schedule for the week using the information below
TOPICS AND AGENDA: 1-Roadmap: Strategic direction: What does it mean for the direction of DDI Lifecycle and DDI 4 work? Take into account any issues raised in TC-189 and TC-203 particularly in terms of relationships between versions How does DDI Codebook fit in? Does it? What needs to be completed to move along the road map Mapping between versions Mapping to other standards (DDI Lifecycle) approach rather than specific mappings, priority mapping to sanity check approach Mapping to other standards (DDI4/Core) - look at output from Dagstuhl sprint Best Practicies document - what else should be included, how can they be made more 'sticky' Technical meets marketing genius session over a couple days would a identify and describe the coolest things to focus on and how to do that at a higher level OUTPUT: Develop some general conscensus for how things need to cohere or align Guidelines for the work to take place in terms of mapping and integration over the week - specific work to take place Strategically where are we going with the production process - What is the starting point for design? UML? Who is doing the design? What are the tools? Who can be involved? [pull together the history of approaches to DDI development - original documents, original DDI4 purpose, development rules for different versions- share prior to meeting - wlt]
2-Basic run through of COGS to answer questions and frame issues Clarification of what is being done during meeting Go through the outputs CSV to XSD to validate approach Are we generating content the way we want to --Break this out --Do the deep dive into the transform process --Verify that everything gets transfered correctly --Have the current schemas been disected to capture everything we want (all the content in a more OO style of presentation) --This is the custom tool that needs the most input We have an existing schema to compare to in 3.3 and 2.5 DDI4 is a different issue and Oliver has been working on it Review rules for trimming as currently prepared Oliver provide a half hour review of his XMI to COGS and round tripping
Production pipeline Binding relationships - COGS is similar to Lion except that the interface is on your computer and using GIT to do the change management and iterate proposed changes for testing
BREAKOUT AREAS: 3-Mapping: Across DDI versions Between DDI and FHIR plus others as appropriate http://hl7.org/fhir/ https://grants.nih.gov/grants/guide/notice-files/NOT-OD-19-127.html How to map - how will we go about this in terms of existing standards and going forward How to express mapping to support understanding and implementation What becomes the point of integration/mapping such as a specific version of DDI 4-Integration: What can be merged in terms of content and what it would look like Where are their conflicts? Structural - implications for future structure of Lifecycle Conceptual - options for dealing with Can we define areas where merger is difficult and general sources of conflict
GOALS: Clearly define issues and approaches M) Structured output on content we have I) what doesn't map What DDI provides that no other standard has Where metadata gets passed on a broader level - what content are you passing/sharing for different purposes Example: what DataVerse uses in from DDI to support their system Interoperability - what needs to be 'sharable' and what
5-Production 5a-COGS Use the document DDI 3.3 Production Validation Document as basis for work Expand during first day meeting to determine scope (in terms of versions) and specific work and assignments Test case for loading GitHub or bitbucket production lines For each version Can we input 3.3 and 2.5 and get out the full content? Make sure we are lossless before we start trimmming Means of comparing input to output 5b-Controlled Vocabulary stuff --From SKOS to other representations 5c-DDI Resolution --Agency ID compared to an arch or DOI handle --Using another resolution system --DNS resolution --Alternative resolution (Dan) --OAI-PMH endpoints and using them as DDI services (imbeded DDI specific versions in OMI-PMH embedding) Need a paper or article about how to do this correctly --Also applies to API packets and what they look like (we have the fragment instance but no documentation on what to put inside of it - best practice minimum viable content etc.
ATTENDEES: Wendy, Dan S., Jon, Flavio, Larry, Jeremy
3.3 entry is in progress No Dan, Jeremy, or Jon next week
Getting 2.5 into COGS will not be a trivial excercise --3.2 required a lot of exception mapping --References needed to be clearly defined --We don't have unique references --2.5 doesn't differentiate between item types and structure so creating documentation or JSON you may need to treat all as structure
One thing we could key off of have the DDILifecycle URN attribte so we could make sure they are used correctly --note that everything deriving from BaseType with GLOBALS has a DDILifecycle URN
May be helpful in cross version mapping or simplification of the Codebook Schema Increasingly persuaded by the argument that we may need to move things into 2.5 to assist in migration
What other pieces of information need to be in the spreadsheet to capture what is needed at the class level
How do you handle docmentation with line ending - see article in modeler-guide use of markdown files Oliver's transformation of DDI4 xmi to COGS
In current version of 3.3 COGS output handled choice and substitution differently because of limitations of non-XML bindings Use a base type and then representation Review deriving and extending from something - there isn't an explanation of inheritance - assumes you have some knowledge of objecting oriented design work and modeling
Helpful to have some How we use COGS in preparation Who is this for? --should people be chucking stuff in? --differentiation between content and modeling --what are the odds of getting lots of contributors?
**Half hour session walking through COGS for clarifications Things on spreadsheet Go through the outputs CSV to XSD to validate approach Are we generating content the way we want to --Break this out --Do the deep dive into the transform process --Verify that everything gets transfered correctly --Have the current schemas been disected to capture everything we want (all the content in a more OO style of presentation) --This is the custom tool that needs the most input We have an existing schema to compare to in 3.3 and 2.5 DDI4 is a different issue and Oliver has been working on it
**Oliver provide a half hour review of his XMI to COGS and round tripping **Get information out early to get some understanding of COGS - there are some quick starts on the documentation page and then the modelers guide **Set up a webinar for walk-through **Does it cover what we want to do with lifecycle?
Controlled Vocabulary stuff --From SKOS to other representations
DDI Resolution --Agency ID compared to an arch or DOI handle --Using another resolution system --DNS resolution --Alternative resolution (Dan) --OAI-PMH endpoints and using them as DDI services (imbeded DDI specific versions in OMI-PMH embedding) Need a paper or article about how to do this correctly --Also applies to API packets and what they look like (we have the fragment instance but no documentation on what to put inside of it - best practice minimum viable content etc.
Roadmap to 3.4 - develop some general conscensus for how things need to cohere or align Best practices guide
TO DO (Wendy): Send out message regarding COGS review - have Dan S and Jon review before then Pull together work plan spreadsheets and post
ATTENDEES: Wendy, Oliver, Flavio, Guillaume
Guillaume's last meeting with TC INSEE has moved production application to 3.3
For TC meeting - production work: Test case for loading GitHub or bitbucket production lines For each version Discuss with Dan Can we input 3.3 and 2.5 and get out the full content? Make sure we are lossless before we start trimmming Review rules for trimming as currently prepared Everyone needs to get up to speed on COGS Means of comparing input to output
Wendy will do a first draft of an A to Z set of activities for moving each version to COGS
ATTENDEES: Larry, Dan S., Jon
DDI 3.3 progress
Dan has made a start and completed / nearly completed classifications work, bits and bobs outstanding, hurdle is DDI-Lifecycle 3640 (Sampling etc), see next item
Jon has made a start on documentation, it will be more user focused, better context and how things fit together
DDI LIFECYCLE 3640 - PowerPoint is a bit confusing to translate into schema changes.
Discussed three options:
a. Not include (discounted as already promised and in Public Release and seen as useful)
b. Only add public review changes and minor updates for missed content (safe option)
c. Update to include rationalisation of process model as per PowerPoint. (concern that major surgery compared to Public Review, would cause problems further down line)
Jon would attempt to take powerpoint and identify changes (not just implied changes) to inform choice between option b and c.
TC Meeting (Minneapolis)
Following on discussion at previous meeting for items
a. Discussion of Roadmap to 3.4 serialisations and content update
b. Mapping between DDI versions - approach as a beginning of taking existing mappings and making them more usable
c. Mapping to other standards (DDI-Lifecycle) - approach rather than specific mappings - might be useful to consider some priority mappings to sanity check approach (e.g. FIHR, schema.org)
d. Mapping to other standards (DDI4/Core) - this would hopefully be an output from Dagstuhl sprint a few week earlier (e.g. GSIM / DCAT / GSBPM are currently on the list)
e. Discussion of Best Practices document- are there more things we should add, can we do something to make them more 'sticky', intersects with discussion on Roadmap above.
ATTENDEES: Wendy, Larry, Dan S., Flavio, Jeremy
See TC-210 for summary of agenda items for cross version commonality and transition
We have a spreadsheet from 2.5 to 3.1, change logs for each DDI-L version, Larry created a generic DDI2 to 4. There is a very generic between 3.2 and DDI4 prototype content. There is a generic model in Colectica that allows input/export between versions.
What gets missed explictly in the 2 to 4 issues of relationships expressed through nesting that need to be made explicit. Example: in 2 you have contact with an affiliation and email but in 3 or 4 you have to create an agent as well as a (2 low level thing apply to the same composit type)
When you go from 3 to 4 there may be more coorespondences
We could do this with COGS model Input of XML schema with a spreadsheet of typing information for all references For 3.3 we are still doing XML editing
The DDI 4 doesn't really follow the roadmap in terms of the Core - there seem to be different opinions on the role of Core and a continuing DDI4 The Core thing is a political thing in that DDI4 is taking too long and so its getting something out but it
The idea of the roadmap was to take the goodness of DDI4 feed it into 3.3 to create a 3.4 How much can we bring into 3.4 Some of it is already in 3.3 (Statistical Classificiation, measures, agents, etc.)
What can be merged what can't and why - can we define these Can we define these areas
Where are things stable DDI 4 could be more RnD work? Publication in 3.4 etc.
Besides doing comparison of Lifecycle and DDI4 We need some means of recording mappings between standards and/or versions of DDI - We have a lot of different products and lots of different mapping - We could use "DDI Integration" model to center the mapping focusing on the portions that are in DDI Integration - Don't need full mapping if it is seen as a bridge
The MRT workplan was focused on integrating with other standards
Agenda item: How would we go about doing that How to go about doing and expressing mappings and itegrations
Defining how we're going to make these mappings is very important
Is there a discussion of "Views" in the MRT? Latest discussion of views is that they will be documentation only Haven't made much progress on integration tools although this may be discussed at Dagstuhl sprint
ATTENDEES: Wendy, Jon, Dan S., Jeremy, Flavio, Larry
AGENDA: TC-208 Generic Agency filing - linked to TC-210 as part of cross-version comparability TC-162, TC-167 follow-up - added xml:space to list of XML attributes - updated guidelines page TC-163 DocumentInformation - discussion
ATTENDEES: Wendy, Jon, Oliver, Larry, Dan S., Jeremy
TC-167 from https://github.com/ddialliance/DDI-TC-167-Discussion/issues/2 (2)Classes of default where there is no inheritance means we'd end up with a mixed set of rules. The simplest approach is to not use defaults. Should not have 2 places it can be defined - it goes where what you are defining is defined Example of decimal representation within a data file (column) - mixing data sets would still have to deal with this but it is clear what column of data is being defined as what https://github.com/ddialliance/DDI-TC-167-Discussion/issues/3 (3)Enforce the rule for default non-usage and the problem of tool support for default becomes moot.
Jon will take a crack at content of TC-203 and TC-189 What needs to get info prior to TC meeting what in the meeting what post-meeting (talk to Barry)
TC-163 What kinds of DDI expressions do we actually have (document, temporary collection of items, passing an object, etc.) We need to flesh this out and provide some good common use cases to help sort out what is needed in various cases and then look at solutions
ZOOM notes: Will continue for up to 24 hours Repeating meeting works the same at GTM in terms of using the link at other times Screen sharing: person needs to request screen with OK by organizer (rather than organizer assigning and then accepting) Local recording of meeting is an option
ACTION: Contact Barry about attending TC meeting for joining discussion of C-L-4 translation and usage; Barry is involved in this with survey work and marketing Goal would be to have Barry involved in TC in this area over the next few years We need to request this inclusion through Exec Co.
ATTENDEES: Wendy, Jon, Larry, Dan S., Oliver, Flavio
Finalize: DDILIFE-3651 RESOLUTION: create XML schema reflecting Classification - Reusable Concept.jpg Note the following Separate exclusion for concept and classification item to support different targets. Notes in diagram refer to XKOS note relationship not DDI Lifecycle notes (document) Dan S. will draft up the XML Change where the level number is Change some of the property names to reflect reference naming conventions
DDILIFE-3657 I have added recommendations for each case identified. PLEASE look at item (3) as we need to determine whether ConditionalResult needs to support multiple languages. Resolved
Additional topics: TC October meeting in Minneapolis --go out next week Creation of a simpler means of filing issues --web form on DDI Alliance (name, contact, version of DDI, description of issue) - respond to filer with link to JIRA issue --concern that it disconnects people from issue tracker - check to see if you can track if not a registered user Encourage people to use JIRA directly but then hook those submitting through form back into JIRA to follow the issue Verify what link is needed to get an issue https://ddi-alliance.atlassian.net/CreateIssue.jspa add project number to link to specify - can do &pid= https://ddi-alliance.atlassian.net/CreateIssue.jspa?pid=10000 DDILIFECYCLE link to Lifecycle source page has list of all - use the TC as the generic
Remaining DDI4 Prototype issues (broad discussion issues) - schedule for next week get additional information on DDI4 position and COGS approach for discussion
ATTENDEES: Wendy, Dan S., Flavio Larry, Oliver
DDILIFE-3656 OK DDILIFE-3657 Review 4 items and comment
DDILIFE-3651 (see issue for more discussion and decisions) 3 versions: not reusable ClassificationItem reusable All reusable
Discussion document describes all use cases - NAICS, Geography
How much reusability is optimal ClassificationItem has an ItemCode as a property
SideNotes from discussion: Concepts - have includes and excludes on Concept Currently this is part of a definition but shuould be able to parse into separate content
using the XKOS vocabulary is helpful in terms of mapping
ATTENDEES: Wendy, Jon, Larry, Flavio, Oliver, Jeremy
AGENDA: 1) RDF vocabularies XKOS published Disco waiting for Achim to release 2) update on Statistical Classifications (status) see DDILIFE-3651 3) GitHub emails - OK to send (fix typos on wiki's and get issues transferred first)
The Technical Committee (TC) needs to tap into your knowledge and experience working with DDI.
A number of issues came up in during the DDI4 Prototype review that we considered as being broader than this one product. Some are technical, some philosophical, some practical. The TC noted that we needed broader input into the discussion than was represented by our membership. So we are bring the discussion to you.
We will be trying a number of approaches, but to start out we have put two issues on GitHub to let you comment, add new issues, and basically, join the discussion wherever and whenever you wish. Each issue will have a description including the goals, background, and some of the related issues noted in TC. Each discussion will be linked back to an issue in our standard issue tracker to ensure they are brought back into the official tracking system.
4) general issue filing Send it to the people who asked for the information and ask for feedback on instructions and process We may need to create a form for the Alliance site which would require transfer but would be doable
5) TC face-to-face (planning/agenda) Issue TC-210 add any additional thoughts as comments Send out email to TC members regarding meeting
ATTENDEES: Wendy, Jon, Larry
AGENDA: review of expanded documentation for methodologies add diagram on methodology relationships Edit current schema documentation FROM:
Metadata regarding the methodologies used concerning data collection, determining the timing and repetition patterns for data collection, and sampling procedures.
Replace with: Methodology covers approaches used for selecting samples, administering surveys or data collection approaches, timing repeated data collection activities, weighting, and quality control.
In DataCollectionMethodology you have.. "A basic structure for describing the methodology used for collecting data."
Repalce documentation in ref=DataCollectionMethodology, element DataCollectionMethodology and complexType DataCollectionMethodologyType with: “Methodologies pertaining to the overall data collection such as primary or secondary data collection, qualitative or quantitative methods, mixed method approaches, GPS capturing methods, methods for collecting data from hard to reach communities, etc. Repeat this element if multiple methodologies are used.”
In DataColletion "This section covers the methodologies, events, data sources, collection instruments and processes which comprise the collection/capture and processing of data. Methodology covers approaches used for selecting samples, administering surveys, timing repeated data collection activities."
Replace last line with: Methodology covers approaches used for selecting samples, administering surveys or data collection approaches, timing repeated data collection activities, weighting, and quality control.
New issue DDILIFE-3654 - Resolved TC-162, TC-167 setting up github for issues and wiki Add information on how to sign in to comment or add issue. Make point of where and how to comment on issues on home page
ATTENDEES: Wendy, Flavio, Dan S., Jeremy, Jon, Larry
DDILIFE-3651 Decoupling Hierarchies from Items and Levels in DDI 3.3 Statistical Classifications Some of the problem was input to COGS and output to XMI - file an issue on that Will continue to try to get information on any changes to GSIM 1.2 and how reuse of CategoryItem may be handled If this takes longer than one-month we can go with what we have.
DDILIFE-3615 DataCollectionMethodologyType seems redundant and disconnected This is a check off item for documentation
DDILIFE-3628 Exposing DDI in more generic search structures like DCAT assumes packaging structures to get at Dublin Core like information Still a future item. Specified some specific steps we will take.
DDILIFE-3619 ResultDetailType's TypeOfResult should be a ref= not name= Need Validation Test for this general problem Validation test for 3.3 would be good but is not a game stopper.
DDILIFE-3644 CLONE - extension of codeBook/stdyDscr/method/dataColl to provide specific code blocks to specify survey settings for data analysis Resolved: addition of CommandCode (o..n) to UsageGuide is sufficent for now
ATTENDEES: Wendy, Flavio, Dan S., Jeremy, Oliver, Larry
RDF VOCABULARIES UPDATE: --Achim has informed me that Disco is ready for a public review. He recommended the minimum review followed by a vote as soon as issues are resolved. --XKOS received only 12 votes all positive. This does not meet the minimum of 2/3rds of the members. There are no rules to cover this situation. Jared will inform the Exec Board that he will approve XKOS for publication pending the approval of the Exec Board.
SCIENTIFIC BOARD MTG: --Scientific Board is proposing a group to review and propose a structure that will improve member input to decisions regarding the direction of scientific activities --A small group is proposed to look at the procedures for review and voting to address new products and issues of non-response ----It was noted that the information needed by voters is not PR blurb but a summary of the review process etc. so that they can determine if due diligence took place. I asked that this happen sooner rather than later as we have a minimum of 3 votes coming up in the next year. --Offered to track process of getting broader input on DDI4 review issues as a potential means of increasing input by members to issues for the Scientific Board. This would mean tracking approach, response, and success of using this approach to increase input. This would be reported back to the group looking at the organization of the Scientific Board.
CODEBOOK: --Talked to staff from ANU working on Codebook and DataVerse activities. There will be new issues filed from this work. --Talked to WorldBank staff regarding new tool. There may be one or two issues filed from this. One definite item needed is a clear best practices for creating extensions and using current contents to capture local focused information (similar to Note, User Defined Key Value pairs etc.) --Several people noted that how and where to file issues is not clear on the web site or to the users. We need to make sure this is clear. Easiest would be for them to file in TC and we could move as needed.
DDI4 REVIEW ISSUES: --Platforms for this activity...suggestions? --GitHub and GitLab for open source wiki and issue tracking ----work in MarkDown plaintext for revision control - can edit directly on the web which is good so we don't have to deal with push etc. ----DDI Alliance already has an organization there --XML would be a good place to start as the DDI4 group is already making internal recommedations
--One of last 2 weeks of October starting 21 or 28 (preference for 21st to avoid Halloween) --Agenda: COGS, finalizing web based agency resolution, cross-version alignment (when to use one and not the other, difference points, frame it out - get some of this done in summer), profiles
COGS MOVE PREP: --Goals ----Is there going to be some type of viewer for state of the model out of COGS ----Every time a change is made in the repository or in your own branch an update will be run generating documentation
https://ddialliance.github.io/ddimodel-web/ ----clarify a wish list of what it does and what we'd like to be able to do --Steps to take ----Prior to meeting ----During/post meeting --Verification of input and output Create a page on TC confluence to start gathering this information
ADDED ISSUE: How to define output formats for DateTime and Numeric Representations - discussed and Dan will file an issue relating to documenting this in 3.3
NEXT WEEK: --3.3 final items --profile of remaining open items on Lifecycle, Codebook, TC
ATTENDEES: Wendy, Jon, Larry, Oliver, Dan S., Jeremy
DDILIFE-3651 table until we get feedback from statistical agencies through Flavio
Went through issues in Prototype Review that require broader input and discussion --The approach will be raised in the Scientific Board meeting --Issues reviewed and clarified were: TC-203, TC-189, TC-167, TC-162
ATTENDEES: Wendy, Dan S., Flavio, Larry, Oliver
We either do this by changing the model or by documenting how to work with this model and create new classification items and add mapping to capture the change.
What is the right approach going forward. GSIM is not going to address the management just remove duplication of attributes (level number, things that are both relationships and attributes will just be relationships)
We should reflect the work of GSIM. GSIM is focusing on the presentation of the Statistical Classification in terms of the rigidity of the hierarchies.
Concerned with all the content within classification item that are sharable across the hierarchies? If all content changes are the same then it is reusable within that context (concept related content, itemCode, etc.). If they change in terms of how they are grouped this should be handled by the the StatisticalClassification. What makes sense for the other properties (what is dependent on the classification what is consistent for multiple uses of the item?)
If it were something other than classifiction item that denotes the combination of a code and concept and then the order and hierarchy is defined by their context. How are the different properties broken up into those that are shared and those that are specific to the classification?
GOAL: --Definately presentation, but clean up and then describe how to manage reuse and classification correspondence (standard varience and versions) but could be used to capture other major reorganizations ----how to create lineage, serious documentation explaining the management part
ACTION: --Flavio will list the properties/attributes that were found by GSIM to be duplicative so we can correct these now --Define which are reusable and which are context (classification) dependent and determine where they should be represented (Wendy will make a first pass) --Statistical classification derived from ....how does this work without an item hierarchy...using same items but different levels (like SIC to NAICS which also brings in the Mexican and Canadian classifications)--ask someone who knows (Flavio will track this down)(StatsCan folks, Franck, Klas)
ATTENDEES: Wendy, Jon, Dan S., Oliver, Larry
AGENDA: Web-based agency id resolution --typo correction; ok'd send out following meeting DDILIFE-3644 Status and 3.3 finalization --3644 - tied to the use for an analysis --3651 - Dan will get updates done to statistical classification so we have updated content to work from and add a "refeerence" name rule check to validation tool Preparation for Members/Scientific Board Meeting --Disposition of Prototype review issues --TC-189, TC-167, TC-162, TC-163 (outlining for presentation to the Scientific Board) --189 ---- No overall official mapping between versions, interoperability,migration from 2 to 3 and beyond ---- Migration should be a major issue for TC both retrospective and how to maintain this going forward (capture as we do the work) ---- Problem of use/abuse of elements over time and the impact on translation ---- How to mitigate the problem in future - description, example, validation tools ---- Possible use of an underlying conceptual model ---- Explain DDI-MF needs to go to the Advisory or DMT to look at intended audience and use target. This should reflect current thinking and be clear in the context of DDI overall ---- Documentation targeted to appropriate user groups, implementers, content people, others? ---- How to interact with GIS and other systems/standardds we do a handshake with ---- Unofficial extension to official library and content. Guidence on how to extend, how to get new content into DDI and how to help interoperability. Can we provide places to manage extensions in a structured way...rules, plug-in points, whats supported in different bindings (xmlconstrain language, rdf full universe), model's custom metadata package, new types as well as extension types (Capture types, pipeline and autogeneration stuff for example). For example in Lifecycle having an expandable custom type or in version 4 that has an annotated identifiable of custom type to ensure correct naming and identification structure.
NEXT WEEK: DDILIFE-3651 Statistical Classifications review TC-167 Default value approach for DDI overall
ATTENDEES: Wendy, Jeremy, Dan S., Oliver, FLavio
review of TC-156, TC-182, TC-193 for closure; --done update on XMI to COGS work; --bring transformations to canonical XMI --movement of content within XMI clarified, transformations need to change in terms of source location (attribute address to child note addresses for example) --includes reworking XMI to COGS which produces XMI back --check on EA XMI to canonical XMI to see if we can get canonical XMI from COGS --Process flow from canonical XMI for 4 is also being done --set up will rely on Bitbucket running pipelines DDI 3.3 remaining work; --Statistical Classification - GSIM 1.1 ----New version of GSIM coming out, some changes are resolving problems in 1.1 ----Issues of reuse of items in multiple hierarchies (no separate relation objects) ----What we have may not be enough for Classification Management ----Send Flavio list of issues related to Statistical Classification ----Get Guillaume and Franck on a call on this in May ----What do we want the 3.3 model to do ----Determine what we can get in 3.3 in terms of what we want it to do ----Primary goal was to publish them in a machine readable form - look at levels and move them from item to classification ----Need to reference an item as it is the right scope/context - need to reference point of use as well as the item --Weighting - check Stos and Ben --schema updates --documentation- and best practices paper --packaging --prep for vote --publication DDI 4 review remaining work --5 broad issues TC-204, TC-205 --Add an issue on use of common properties such as "typeOfXxx" across versions --Pull these together as a broader discussion topic in TC in terms of cross-version comparability and transversing across versions
TO DO: Get Statistical Classification sorted out in May for 3.3 Contact Stos and Ben regarding weighting
ATTENDEES: Wendy, Larry, Dan S., Jeremy, Oliver, Flavio
AGENDA: 1) DDILIFE-3616 2) Post_Prototype disposition (14 items) - DONE 3) framing discussions for TC prototype review issues -- One approach is to outline issues, identify communities/individuals, gather input, organize focused discussions 4)Send out an invitation to discuss web based resolution options and get use cases. Dan S. will write up introductory information.
ATTENDEES: Wendy, Dan S., Jeremy, Oliver, Flavio (Larry provided comments via email)
TC completed triage of Prototype_Review issues and moved 19 issues to DMT for resolution. Comments were added as needed. Of the other issues not previously moved, 3 issues required no changes as the content already existed, 3 issues were direct fixes (typos, minor addition, etc.), and 5 issues require broader discussion and response. TC will organize these issues and set up a process for obtaining community input and discussion.
Note that there are still 14 outstanding issues identified by the TC during the technical review process which need attention within the context of the Prototype Review.
ATTENDEES: Wendy, Dan S., Jeremy, Oliver, Larry, Flavio
AGENDA: Prototype Review issues: --Triage goals --those that can be handled by TC or that can be made declaritive --clarification --Approaches for disposal --Identify where the fix is clear, state correction to implement --Identify issues that are DDI4 specific modeling issues - comment where needed, return to DMT --Identify broader cross-DDI issues - frame issue, organize discussion among stakeholders to gain input and inform decision. Implementation in DDI4 would then be coordinated with MRT group (who would have been part of discussion) --Follow-up --TC will track action on Prototype_Review issues and report back to SB
Start working through issues Triage work done on: TC-156, TC-161 through TC-169
ATTENDEES: Wendy, Jon, Dan S., Jeremy, Oliver, Larry
DDILIFE-3615 DDILIFE-3616 TC Roadmap - finalize and prepare for passing on to Scientific Board and Exec --web based resolution should be added - Dan S. will provide text 2019/2020 budget requests --this should include a discussion of where we are with COGS, what needs to be done in terms of DDILIFE, DDI Codebook, DDI4 --a discussion about things in the road map - a face-to-face on Codebook 2.6 and other roadmap issues / alignment with GSIM / RDF serialization that can mix and match --draft something about above --would it be worth suggesting where it would be - who needs to be there and determine where is the most economically feasible --Current work is all concluding up around July August so that might be good [August bad, Dagstuhl 9/30-10/5, October looks good so far hold 3rd week 14-18 or 21-25?] --If 2.5 might want to get Mehmood, good to have Olof and Johan
UPDATES: --Where we are with DDILIFE entry for updates and documentation - entry this weekend --XKOS update - ready for vote --Disco - sometime in spring - will it need a public review --Responses regarding CV usage - SKOS, JSON both mentioned as well language filter for HTML
DDILIFE-3644 read the example and be prepared to discuss
Add command code but then make a future issue to explore specifications for different types of weighting to help generalize to type. Document use of UserAttributes to capture some of this information. Look at this as we review the related issue in for 2.5 - we want to be sure these align One concern is that if we get too specific about one type of weighting it may become too specific in one direction only What is stated for STATA could be machine processed to other language Use example of translation of how to create Key/Value pairs Resurrect suggested CV from Weighting group and Sampling Group and pass on to CV to create and support new content
AGENDA: Wendy, Jon, Oliver, Larry, Flavio, Dan S.
DDI4 Prototype Triage pass --During full Triage we need to identify which issues have broad implications and make sure we have a list of "types" for triage that allow for more sorting of issues to be addressed. Do this before end of March. --Move 4 identified issues to MRT --Assign TC-189 to Jon to start forming a protocal for looking at these broader issues
TC Roadmap document: Basically the roadmap reiterates that DDI-C and DDI-L should not be forgotten. We have new users of DDI-C as well as existing systems that are invested in DDI-C and DDI-L. This is an updated version of the 2017 document. We feel it would be helpful for the Scientific Board and Executive Committee to see where these products are, how they fit into the mix, and the type of continued support required. This also seems to be a means of re-engaging membership who are concerned with current systems.
There were some questions raised about the second to last paragaph in the Lifecycle section in terms of short- and long-term goals of MRT in regard to coverage of DDI4. Larry will provide some clarifying language.
We aren't addressing all things in DDI4. There is an issue of being able to move featues from DDI4 development into the 3.x series to make them available to users. This would be done in response to requests from the community.
One specific area is Classification Management. Flavio will look at what has been added to 3.3 (GSIM model) to see how it lines up with DDI4 and XKOS. The goal of this work now will be to prepare a best practices guide re: XKOS and Classification Management section.
In general we need to have mapping information built into the modeling. We need to be aware of interations and crosswalks in mapping. Overall we need to have a continuing focus on not only how a version of DDI interacts with other standards but how it interacts with other versions/products of DDI.
DDILIFE-3630 decision on supplement DDILIFE-3648 decision DDILIFE-3615 review new document for discussion DDILIFE-3616 Dan S.? [no new info] DDILIFE-3644 example and Universe question --look at example for next week --The proposal seemed a bit to specific to a specific type of sampling (might be good to have Dan G. look at it also) --There are parameters of the sampling that need to be identified beyond just the weight --should weights point to variables? (currently variables point to weights and the weight to its generation and usage
NEXT WEEK: DDILIFE-3644 (DDI perspective)
ATTENDING: Wendy, Jon, Dan S., Jeremy, Larry, Oliver, Taina, Sanda
NOTE: post call email discussion added at end to clarify these notes
- Does the DDI Alliance want to publish the translations of the CVs as well as the source CVs? That would require changes in the website. If not, I think there should be a reference to the CESSDA Vocabulary Service CV which has the translations. And even if yes, a reference anyway? DISCUSSION: There is only one reference to the CV. The SKOS is already merged. The tools had separate versioning for languages. How to deal with versioning. The point is what we want to publish for the alliance. We want to publish multi-language where available. We are publishing different formats transformed from SKOS. We can include a reference to CESSDA Vocabulary Service CV in the documentation.
- On the DDI Alliance website, the different location URIs of different vocabulary formats take too much space. Can this be amended, for example, by presenting a button for each format? DISCUSSION: Which page are you referring to? html - language or format
- URN: at present the ‘Canonical URI of this version’ element (=urn of the version) does not have a language tag. The IT people for the CESSDA tool say they would like also the source vocabulary have a language tag. This would mean that the urn of the English CV version would be: urn:ddi:int.ddi.cv:TypeOfInstrument:2.0-en. Is this OK? DISCUSSION: Good news is that the developer said in today that it is no problem for the system not to have language tags in the urns.
But there are a couple of other perspectives: · An identifier should not have any semantic on which a system relies on. The fields agency and version are exceptions; they have a very specific definition( according ISO 11179). · Common practice to deal with multiple languages of a resource (HTTP header, HTTP negotation, multiple languages in on RDF resource) · Agreed way in the DDI Alliance context regarding CVs with multiple languages
Display issue in terms of viewing the full list (in single languages) or a term (in multiple languages)
The purpose of the html is viewing and browsing so we have some flexibility in how to approach the display. We are publishing the standard not a tool to explore the standard so we need to make it clear to support users, but not necessarily creating the ultimate display. The primary aim is to publish the CV's in usable formats (SKOS, Genericode, CodeList, JSON, this should be decided by the user community, is there a codelist layout for Codebook as well as Lifecycle?). Need to come up with some box and arrow drawings of how we'd like this displayed.
- Usage information may be presented differently, can the publication pipeline just publish whatever is there in the CVs in the CESSDA tool? Module/attribute/element names will be there with a link to the specification, as well as element numbers for DDI2 but no definitions of the element as now in DDI website. DISCUSSION: It's useful but it would not be synched with the latest documentation. It should pull the documentation from the DDI publication if possible. This could be done in transforming SKOS to html with full language and content and then select box for display (language or term by term)
- The tool will automatically assign the publication year as the copyright year. For all CVs published from the CESSDA tool, it will be the current year, be the CV old or new.
DISCUSSION: its ok
- Urn resolver planned for DDI? DISCUSSION: The currently situation. DDI agency administration tool is ready to resolve to the agency supported internal resolver. We need to check with ICPSR but don't think there should be a problem there. This still needs to be filed in order to achieve the first step, resolving urn:ddi to the agency administration tool. Current workaround is to recommend us of both urn AND the url if using thse CVs
The point is that the CV's have a citation and the citation has the URN not the URL
ACTION: TC needs to determine what people are using through the broader user community (ddi-users, ddi-members, ddi-scientific board, ddi-srg) - send out soon
PRODUCTION: When to publish changes (partiularly language additions) Production process - review, creation of bindings, input to DDI documentation (how to capture intermittent updates to documentation), approval, announcement Prepared before summer meeting, get it on the agenda
License of translations - will be the same as the source ccBy4
FROM CV: Functionality of the CV site on DDI Alliance in a more formal way
In practical terms what triggers the output from the CESSDA system regarding a change in a CV. Harvest system to check for changes and push out a SKOS to the DDI repository for entry to the DDI production pipeline. DDI internal production system, to web. Clarify transfer points and actions and who does what. (Who being an agent - machine, person, etc.) Change logs for individual CVs. Clarify requirements for display, usage, versioning differences, and clarify what needs to change and priorities for those changes.
POST CALL EMAIL DISCUSSION:
Q: If I understood correctly, DDI Alliance will want to version the whole package with all language versions? If there is change in any language versions, the whole package version changes. I can see this is useful if, for example, the CV is used as a multilingual filter in a search interface. RESPONSE: If I understood Oliver correctly we could get a merged language output in SKOS from the CESSDA tool. If not it is simply an additional step in the production pipeline to create a merged version from the individual language versions prior to transformation to other formats and publication. Oliver should have this in his workflow.
Q: The CESSDA tool versions each language separately, though the translations have a three-digit version number linked to the source as the first two digits are the source version number (i.e. 2.0.2 means the second TL version of the source version 2.0). This is useful for users who are using the CV in their metadata. They are only interested in when and what has been changed in their language version to see what needs to be changed in their legacy metadata and in their systems where the CV is implemented. Q: The issue maybe to discuss is how to produce the whole package versioning, in the merged SKOS of the CESSDA tool, or in the DDI publication pipeline? But this is technical, so out of my expertise. RESPONSE: Basically we will publish a SKOS instance that contains all languages and therefore have a single version number. How the version number is defined can be up to us. So yes we could say that the major version doesn't change unless the terms change or there is a significant correction in the intellectual content of a definition. The reference could be to the major version and the user would get the most recent sub-version of that. Some people do sub-versions simply using dates and internal change log with a date, change, and reason for change. For example: 2019-03-15, Term X definition (en) Aggregaetion changed to Aggregation, Corrected misspelling OR 2019-03-15 Added labels and definitions in German (de), added translation option. In short, it is up to us to determine the versioning rules and make them clear to users. In making a decision on this the primary issue should be user needs. What changes actually have an impact on the user and the value of stability in the version numbers.
Q: The link to the Unesco vocabularies was just an example of the one typical html display of multilingual vocabularies. Based on open source publication platform skosmos. It displays the whole language version on the left and term-specific display on the right. In term-specific display all language versions of a specific term are displayed. What is lacking, I think, is the possibility to click on any language term on the right and change language that way. RESPONSE: will add this to minutes to clarify
Q: Regarding the citation, the citation in the CESSDA tool has both URN and URL. Did we conclude that in CESSDA tool the citation URL is to that tool, and in DDI Alliance website the URL in citation is to the DDI Alliance webpage? That would need to be taken into account in the DDI publication pipeline. RESPONSE: Yes, the DDI publication will provide DDI URN and URL (probably to the SKOS version as the source)...that can be decided. It should line up with the target of the resolved URN. Actually, since we have multiple formats we will need to consider just WHAT the URN is landing on. This may result in reorganization of the web content.
RESPONSE: The primary concerns of the call with TC was to make sure we had identified all the technical things that needed to be decided and completed. In particular to identify interaction points with the CESSDA tool, the production pipeline, CV/TC, and ICPSR/Web page. I will add a line to the minutes that the UNESCO link was simply and example
ATTENDEES: Wendy, Jon, Dan S., Jeremy, Larry, Guillaume
DDILIFE remaining items (x'd items have been dealt with see issues for comments): 3614 3615 3616 review with 3615 3619 waiting for validator 3647 x 3645 x 3646 x 3625 x 3641 x
Put on later agenda: documentation production
ATTENDEES: Wendy, Jon, Dan S., Larry, Guillaume GUESTS: Stas Kolenikov, Ben Hilland (DDILIFE-3644)
DDILIFE-3544 Research Organization working on Government Contracts (Federal Agencies) STATA Code developer Publication of survey design features in metadata (ICPSR data) Processers need information on how to process information for accurate use of weights, estimations, etc. based on survey design, sampling design, variety of weights. Improve content to support use of complex weighting content.
Do you think there is a level of adoption in the community your working with? Hard to say. People have become more aware of the need to reflect impact of sampling approaches on weighting the use of weights, therefore credibility of responses.
Alterntives: list a variable that states, strata, cluster, etc. The variable can know about itself by defining its role. Do you expect DDI 15 years from now? yes..its been in use for 20+ years already. Parse R or Stata statement if avialable and self-identification of variable roles Are we going go proprietorial around current issues? Weights, Cluster, Strata and Calibration variables accounting for point and variance estimates. For some time R could do calibrations properly now STATA does (in past year). Also use of Replicate weight. If you could have command code that would provide a flexible specification that would address. Ben: Would it help if we tried writing an XML stucture to describe this? We should probably start off being ambicious in describing these sorts of issues.
Why isn't this already addressed in 3.3? Role of a variable can be stated, and preferred weights? Should there be a means of fully structuring how to get standard errors, replicate weights, common weights. A clear link between variable and specific survey design and weights.
Use the more complex example with to do an example to see what could be captured currently or with the additional commandcode content.
Roper poll data (replicate weights)
The specification of the cluster, strata, and weight variables as a set have a purpose and there are different variables used for different purposes. At the same time you can say a variable has an inate grouping of variables in terms of their case (person, household, etc.).
May be a benefit in
ACTIONS: 1-Stas and Ben try writing an XML stucture to describe 2-Add weighting information and variable content to issue so they don't have to wade through 3-ANES example and CPS example - see where they interact with the sampling process 4-Follow up with Stas and Ben in 2 weeks to see where we are
ATTENDEES: Wendy, Jon, Dan S., Larry, Guillaume
DDILIFE-3640 Sampling - ok'd for finalization any remaining 3.3 issues - reviewed to make sure what has not been addressed by 3640 and scheduling 2019 workplan - reviewed and updated
DDI Agency Registry update: New version is now up and running - more information for approvers Cross platform (runs on Linex) - Jared is reviewing it Now can have your own user account that can manage agencies rather than agency The TC can think about means of resolving services for agencies over the web in addition to over DNS --for discussion this year (added to workplan) Offical URN namespace registration (added to workplan)
ATTENDEES: Wendy, Jon, Dan S., Jeremy, Olvier, Guillaume
DDILIFE-3640 --Put in link from DevelopmentImplementation to DevelomentProcess - got lost in shuffle
--Makes sense that its restricted right now as this is the bounded problem we are trying to solve. Leave at Question, Measurement, Instrument, ControlConstruct reference stack. Does anything else need this within the context of Data Capture development? Seems to cover the major development components.
Sampling: --Use references to UnitType and Universe where applicable --Complete example and send out next week following final review by TC
DDILIFE-3644 --Contact Stas and Ben about availability for call next week. See if Larry will be available
TC Agenda for 2019 To Do list: --Codebook work to do (good summer project) --Prototype dispensation of review comments (end of March) --INSEE might get something in --Oliver is checking on GESIS --Darrin?? --Publish 3.3 (end of March) --Move to COGS (Spring/Summer) --New system for editing controlled vocabularies - need process to move SKOS to DDI outputs - work with CV group, get a meeting with the CV group on this to speced out and get a time line set up
Wendy will draft work plan and send to Jon for additional discussion and then will bring to TC
ATTENDEES: Wendy, Larry, Oliver, Jon, Dan S., Jeremy
3626 - simplify name to ReviewProcess but "process" seems to imply the ApprovalReview
3640 - got through Data Capture Development section. TC group will look in more detail. Need to get feedback from people doing this work to see if it sufficient and/or overkill. Send out ppt and example after providing the descriptive (non-actionable) implementation.
Look at new issue on weighing for next week
ATTENDEES: Wendy, Jon, Larry, Dan S., Jeremy (Oliver had connection problems)
3616, 3617, 3626 and 3643
Jon will take a bit more of look at 3626 but I think its pretty well resolved
Dan S. is providing a short list of items in 3616 that need to be addressed now
Wendy reviewing the final list of changes for 3617
ATTENDEES Wendy, Jon, Dan S., Jeremy, Larry
Reviewed remaining open issues in DDIL_3.3_review Remaining Methodology issue in Master Issue are waiting on examples to be completed - all should be resolvable at that point DDILIFE-3617 was discussed; members should review the spreadsheet and identify where extension should be replaced by the creation of a element containing the original extension base and the added content DDILIFE-3616 review Identifiables for consistency of practice and note if they should possibly be Versionable or a property instead of Identifiable DDILIFE-3614 propose revision of documentation which was writen before UnitType was added. Verify with Guillaume concerning alignment with GSIM (Larry will file a related issue for DDI4 in TC)
Note that resolved issues with clear changes listed in issue have been sent to Dan S. for entry Remaining items need specifications provided and will then be sent to Dan S. for entry
ATTENDEES: Wendy, Dan S., Jeremy, Larry
DDILIFE-3615 DataCollectionMethodologyType seems redundant and disconnected DDILIFE-3616 Why are all the content types within MethodologyType identifiable? Revision of Sampling and Development activities to make use of control constructs - what new Acts would this require DDILIFE-3605 DDILIFE-3620 NOTE: Guillaume has added a number of comments to 3620 that were not reviewed by TC prior to the following discussion
(these notes also appear on the MASTER ISSUE: methodology review)
--Having the multiple ways of setting up ordering is much cleaner using existing control constructs creating more acts --In parameters and bindings are we constrained on what we can pass through processes that have known inputs and outputs --Is the question that there are other forms of metadata that may be passed through and we don't know such as embargo information on development work --If we want to do sampling stratification and passing if we need further Acts they can be added in later revisions
TO DO: Write it up and run some examples to see if there are roadblocks
AGREEMENTS: --Change to Data Capture Development --Do example of both question and measure --Use CPS and PUMS for sampling plans
Issues to clarify in documentation: --Use Sample is the result of a Sampling activity. Need to document the plan, the execution (allows docuementation of specific iteration variance, specific activities) and the result (sample). --UnitType, Universe, Population difference and which is associated with each (Frame, plan, event, result) --Specification of Sampling Methdologies. The sampling plan is more of a methodology than the sampling (prodedure/event/action).
ATTENDEES: Wendy, Jon, Dan S., Jeremy
updated roadmap for coming year - Jon has updated version of mid-2017 document - If "DDI4" is not a continuation in terms of coverage then it should be a separate product starting at a version 1.0 - The point is the discussion needs to take place in the broader community and it needs to cover both the definition of DDI4 and support for current usage - One issue that will come up would be multiple RDF ontologies that will need to be namespaced so that there is not a clash between DISCO, DDI-L and a DDI4 (Research View) - Circulate to SRG list ask for comments and a deadline (4 weeks) - List of tasks for the 3.3 release -- List of issue numbers that have resolutions for Dan
methodology organization and clarification (includes new content) - Methodology is mostly textual content - Do we need all the new types and if so should they be complex data types or versionable - Identifiable is difficult to handle - The number of types has increased in 3.2 when IdentifiedStructuredStrings became identiable bundles of typeOf and description - schedule for next week
Next week DDILIFE-3615 DataCollectionMethodologyType seems redundant and disconnected DDILIFE-3616 Why are all the content types within MethodologyType identifiable? Revision of Sampling and Development activities to make use of control constructs - what new Acts would this require
TO DO: Post spreadsheet of issues Post spreadsheet of methodology etc. Get Dan list of issue numbers with resolutions Circulate the roadmap
ATTENDEES: Wendy, Larry, Dan S., Gulliame, Jeremy
DDILIFE-3630 Added comment - find and add examples in JIRA DDILIFE-3628 discussion needed - assigned to FUTURE following resolved: DDILIFE-3634 Background information and recommendation added to ssue DDILIFE-3631 typo - fix it DDILIFE-3629 typo - fix it DDILIFE-3628 discussion needed - assigned to FUTURE DDILIFE-3632
More open discussion of future broader issues: --Add new issue for DescribableType usage DDILIFE-3641 --Another broad change would be how the groupings are done - another sweeping change in terms of order and specifics etc. --Switch just to item by item definitions rather than choices (inline or by reference) --If we are talking about broad changes in future versions these would be main ones
CV related discussion: We need to get back to the CV talk about publication process etc. The work on the management tool is progressing to a point where we need to restart these discussions. Put on work schedule for coming year.
A broader question for DDI is as we get beyond questionnaire should the CV group be broader to reflect broader coverage and interest (non-questionnaire/survey things). How these get into DDI CVs or acknowledged by DDI as recommended for use.
ATTENDEES: Wendy, Jon, Dan S, Jeremy, Larry, Johan
DDILIFE-3615 DataCollection/MethodologyType seems redundant and disconnected DDILIFE-3616 Why are all the content types within MethodologyType identifiable? DDILIFE-3621 There are many interlinked references in the new sampling and development area DDILIFE-3605 Questionnaire development should probably use the work flow capabilities DDILIFE-3620 DevelopmentProcessSteps seem to be reinventing the process model DDILIFE-3622 How does questionnaire development relate to Instrument
TODO: make a master issue, do background work to visualize what we currently have, intent etc. (see notes on 3622)
DanS: make a list of identifiables (3621, 3616) for next week
Next Week: DDILIFE-3630 DDILIFE-3634 DDILIFE-3631 DDILIFE-3629 DDILIFE-3628 DDILIFE-3632
January meeting: DDILIFE-3615 DDILIFE-3616 DDILIFE-3621 DDILIFE-3605 DDILIFE-3620 DDILIFE-3622 DDILIFE-3617
ATTENDEES: Wendy, Jon, Oliver, Achim, Johan, Jeremy, Larry, Dan S AGENDA 3.3 to COGS -image from yesterday 3.3 production flow using COGS -recommendation to use the same production pipeline for all versions in COGS using a canonical XMI base for production of representations XSD format/style -Pure GardenOfEden -Nested XML -Imposed order -Single/Multiple Schema XMI output style to support any of the versions -Handling Packages in XMI and imput process Where is the best place to validate incoming CSV files -Possibility of a simple tool like Dan S. creted for the schemas -Validation within COGS of csv
DISCUSSION: Decision for production line of DDI-L and DDI-C does not need to happen at once Publication of 3.3 is not dependent upon getting everything working in COGS We now have DocFlex output from 3.3
COGS maintains element and attribute and supports substitution group of any object derived from the complex abstract (substitution head) COGS also retains order information so that publishers who need to specify order will be able to do so
COGS has a lot of internal validation testing Validation points for testing -validation of CSV prior to build validating against model construction rules Webhooks for new commits are made when with the run build script with validation testing Correct or failed are reported with link to build run If needed can run these on a forked branch but this will require a hook up...directions will be added DDI only works on pull request to these Could also run on own machine
Oliver has created an XSL transformation to COGS csv based on canonical XMI COGS XMI output is currently EA which we should be able to run through Achim's transforamtion to canonical to see where problems may arise
There are currently abstract classes in CSVs which must have some sobstitution class. Item-to-Item and Complex-to-Complex substitution groups are available
MEETINGS IN DECEMBER 6 CANCELLED - EDDI 13 MTG with most memebers - Set up review of new SDI content (sampling, development, weighting) 20 Probably will meet 27 Probably will not meet
COGS has a lot of internal validation testing
ATTENDEES: Wendy, Larry, Dan G., Dan S., Oliver, Jeremy, Flavio, Jon, Johan
Primary agenda items: Triage on Post_Prototype and Prototype_Review issues for EDDI Sprint - DONE Key items for TC work at EDDI - send recommendations via email and we'll work that way to organize the agenda
No meeting 2018-11-22 Call will be held during sprint 2018-11-29
Post-Prototype issues and sprint Triage of Prototype_Review Suggestion: review and move Post-Prototype to MT - triage next week to move Review this week and assign to MT next week
Revisit DDILIFE-3625 Dan S. will provide more specifics on issues this change would address and requirements for the change. Issue was reopened for further discussion. The issue discussed last week regarding documentation changes is being moved to a new issue and placed under discussion as some concerns were raised. DDILIFE-3633 Decision made
DDILIFE-3633 Decision made
DDILIFE-3626 Discussed where this would link and determined that it might be more useful to have a more generic Administrative Review that would cover IRB (Ethical), OMB question review, NIH variable review, etc. Dan S and Larry will gather more information on these additional review types. The goal of this information is to track that reviews of specific items (project, question, variable, etc.) has taken place and approval obtained. It is not a review management system. Still collecting background information approach of a more generic structure approved
DDILIFE-3624 Clarification of original request and some discussion of suggestion to do all inclusion of versionalbe and maintainable objects by reference. This may be too large a change for 3.3 but Wendy will look into level of change required. It may be just a best practice recommendation (use of reference) to prepare for future versions. Specific issue of Instruction property in questions decided Broader issue of versionable maintainable inclusion
Next week: Triage of exisiting Post_Prototype and Prototype_Review
3626, 3616, 3617, 3629
ATTENDEE: Wendy, Jeremy, Dan S., Larry
AGENDA: Revisit DDILIFE-3625 Dan S. will provide more specifics on issues this change would address and requirements for the change. Issue was reopened for further discussion. The issue discussed last week regarding documentation changes is being moved to a new issue and placed under discussion as some concerns were raised. DDILIFE-3633
DDILIFE-3626 Discussed where this would link and determined that it might be more useful to have a more generic Administrative Review that would cover IRB (Ethical), OMB question review, NIH variable review, etc. Dan S and Larry will gather more information on these additional review types. The goal of this information is to track that reviews of specific items (project, question, variable, etc.) has taken place and approval obtained. It is not a review management system.
DDILIFE-3624 Clarification of original request and some discussion of suggestion to do all inclusion of versionalbe and maintainable objects by reference. This may be too large a change for 3.3 but Wendy will look into level of change required. It may be just a best practice recommendation (use of reference) to prepare for future versions.
Next week: Come back to 3624, 3623, and 3625 if background work is done 3616, 3617, 3629
Availability for meetings prior to EDDI: Not Thanksgiving - all Americans Will have the TC call during sprint week
ATTENDEES: Wendy, Jon, Dan S., Jeremy, Larry, Guillaume
Started work on 3.3 review items DDILIFE-3606 DDILIFE-3607 DDILIFE-3608 DDILIFE-3609 DDILIFE-3610 DDILIFE-3611 DDILIFE-3612 DDILIFE-3613 DDILIFE-3614 DDILIFE-3618 DDILIFE-3619
ATTENDEES: Wendy, Jon, Oliver, Dan S., Jeremy, Johan, Jay, Achim, Kelly
Scorecard -tc-8 TC review document tc-8 list for page tc-136
remove "read before reviewing"?
ATTENDEES: Wendy, Jon, Dan G. Flavio, Jay, Jeremy, Oliver, Kelly, Larry
TC-128 Flavio's comments on review of sampling methodology Read to see if the documentation works to explain the methodology pattern. What wasn't clear was the difference between Algorithm and Process at differet levels of abstraction. Is there a clear cut between the two? Will people use them in the same way or as intended? Documentation could be improved to convey this. Example to explain difference was difficult for someone who is not as mathematically literate to understand. Did Flavio get it right? Are you questioning the model or the document? Is the alogrithm needed? What is its purpose - high level description of theory vs. practice where algorithm is the theory. What actions do we take? On documentation is the methodology pattern explaines the alogirthm but not really the process. Questions on the structure of the model should be filed as issues and labeled post-prototype. Jay ran into the issue Flavio talked about regarding lack of structure. Used structured text to provide steps. Kelly will put in some text for Dan to check.
TC-8 DDI Prototype TC Review Comments and Comparison spreadsheet
TC-136 What if any guidance do we want to provide the reviewer? - See issue
ATTENDEES: Wendy, Dan S., Jay, Johan, Larry, Jeremy
TC-8 Review of Prototype by TC document items 5-8
Wendy will summarize and draft documents for publication with prototype
ATTENDEES: Wendy, Dan S., Flavio, Jay, Johan, Larry, Dan G., Oliver
Review of documents from TC for Prototype Review:
Relation between Post-Prototype issues and Lessons learned section should be integrated for clarity
Another area we need to look at is what to use as a modeling environment. We need to relook at EA and modeling environments.
We've included the rules that we're using to go to bindings
Some bullets may need further clarifications.
Need to review sub-set used of UML and how they are represented in bindings.
Post-Prototype we need to have a real discussion of what UML features we use
Change ComplexDataTypes to StructuredDataTypes and double check for any other name changes
Went through items 1-4 on Review of Model by TC
NEXT WEEK: Review of Model by TC items 5-8
ATTENDEES: Wendy, Dan S., Flavio, Jay, Jeremy, Johan, Kelly, Larry, Jon, Dan G.
Validation Checks Modeling Rules We need to provide a framework for review - what TC has done, what is still open, what is the thinking Goal is to have some description and where this stands From a developers perspective - there are issue Need to collect TC member perspecitives
ACTIONS: Move documents to issues (attach or link to google doc) During the next two weeks enter positions on the google docs for discussion next meeting (2018-09-06)
ATTENDEES: Wendy, Dan S, Flavio, Jay, Jeremy, Dan G, Johan, Jon, Kelly, Larry
TC-9 through TC-14 contain comments regarding additional documents that need to be prepared particularly specific TC review documents and best practices. If there is an open one that you can draft, add your name in the appropriate comment.
Wendy will start a set of TC comment drafts in each area to encompass current work. See TC-9 through TC-14 for links as they are added
Documentation production is still a problem. Jon will follow up with Oliver and Achim regarding doc flex
ATTENDEES: Wendy, Dan S., Jay, Larry, Oliver, Flavio, Jeremy, Dan G.
TC-153 Documentation content
TC-125/TC-152 results of statistics review require changes
TC-151 Need addition to ConceptualComponent to do example
TC-150 InstanceVariable measure (target Population) is required MAKE OPTIONAL It has not been used in some of the examples. We need to either change measures (target Population) from 1..1 to 0..1 or go back to the example authors for additional content information Is this the differnce between RepresentedVariable and InstanceVariable. Both specialize Concept. Do you always know the Population when you measure? You might know it in a less structured or formal way. You require that formalization to have the Population object. So in overwriting the target it violates the rule of constraining to a sub-type and Population is not a sub-type of Universe. Is this predicated on the fact that the relation is called Measures in each case. Call it something else and Instance Variable ends up with both Universe and Population. InstanceVariable can't be reused from Data set to Data set (for examples in repeated samples). Make a note that we need to think about how to rename or create extension of Population from Universe. Recommendation for a change in RepresentedVariable from Measures to DrawsFrom (Universe). Remove narrowsUniverse from Population. DID NOT REMOVE from population as the inheritance is structural not content
TC-143 Change in Process Pattern and related realizations completed. Specifically this has resulted in some changes to relationship "executes" in Workflows as this is now inherited from WorkflowControlStep (0..n) and is constrained to 0..1 in conditionalControlStep. Note that ElseIfAction also now uses executes as the relation name as conforming to consistency between name and target pairs where resonable.
TC-112 This has been filled in, any changes?
TC-91 Examples have been updated and required content added with the following exceptions: measure in InstanceVariable until TC-150 is resolved properties that have default values in the UML which have not yet been programmed for a standard translation into the bindings Question about incomplete content (see exampleIssues.txt on issue) - are these intended to be examples of best practice or is the information simply not yet known?
Work with Jay to add content
TC-136 Issues to document in Prototype TC evaluation: XML limitations: transformation of primitives xs:decimal, xs:float, xs:double, xs:date, xs:datetime; transfromation of default values and regular expressions; need for secondary validation to enforce cardinality constraints Relationship of Statistical Classification and XKOS DDI-C, DDI-L, DDI 4, GSIM, SDMX, ISO-19115, ISO-11179 relationships Ability of each Functional View to support stated use cases Use of Functional View (separately published XML schema) to address requirements for focused application guidence RDF vocabulary - overall statement Use of UML modeling options and translation rules to bindings Statistical Summary (TC-125) Formal documentation limitations, goals
ATTENDEES: Wendy, Dan S., Jeremy, Johan, Kelly, Larry, Achim
TC-98 Larry will pursue wording with Jay and Dan G. - updated wording, work is being done on a supporting document for review TC-128 Dan G. will provide draft tomorrow - unable to complete (Dan G. is now on vacation) TC-144 Wendy will enter - done TC-143 Wendy will enter - done TC-112 It's on Kelly's to do list - today
TC-9 - TC-128 (Dan G. working on this), TC-125 statistics packages - wlt is reviewing and writing up TC-10 - Only outstanding is the document describing use - Achim is working on it TC-11 - Working on Prototype workarounds - default values, translation of UML primitive Real to XML; paper on need for secondary validataion TC-12 - Achim has completes an XML to RDF translation and is rendering examples TC-13 - Round trip - see TC-12; statement regarding goals TC-14 - Documentation - today Make sure there is a discussion of Packages vs Function View in the high level
Extend 3.3 review for 2 weeks through the 20th - email to Barry and Jared Follow up on Oxygen documentation for 3.3 with Jon
ATTENDEES: Wendy, Jeremy, Larry, Dan G., Flavio, Kelly, Achim, Eric
First half of meeting (many of these are just touching base and getting estimate for completion): TC-144 resolved TC-143 resolved TC-128 eta tomorrow TC-125 updated TC-113 updated TC-98 updated TC-91 updated TC-43 updated
Translation decisions: Works with canonical version of XMI Views and model presentation - captures the XMI inheritance faithfully Enumerations (value set) Various views of content (FHIR, ShEx, OWL viewed in Protoge)
Choices: Document describes choices made Transformation rules Existence content makefile DDI specific (goal) Arictecture
parse XMI For each class and property remove realizes For each package ending with Pattern remove xsd datatypes were inserted as LION doesnt' really export this
Two properties with missing elements because they were in one of the pattern packages - send to Wendy
UML model that is DDI specific transformation Turns into ShExJ (Jason representation)
Package doesn't have semantics in RDF but is handy for finding stull Used just for organizationally purposes.
Handy for organizational purposes, or that something is an enumeration (semantic information) otherwise navigational.
Who is the audience for the RDF material. Implementors usable but for general public need more high level description. Current audience should be fluent in semanitic web technologies. More of a review of the technical representation, is it good enough, etc.
Need a statement of what this is a review of. What is the purpose, how it is useful, etc.
An example in RDF of Australian National Election Study (parallel with XML) - can limit to a single instance of different types of variables. Wendy will provide the XML.
See FHIR example
Not worth writing the translation right now. Might work on it following.
There was an example from the last Dagstuhl meeting (Achim will find)
ACTION ITEMS: TC-98 Larry will pursue wording with Jay and Dan G. TC-128 Dan G. will provide draft tomorrow TC-144 Wendy will enter TC-143 Wendy will enter TC-112 It's on Kelly's to do list RDF - Achim will locate example from Dagstuhl - Wendy will provide Eric with AES XML and PDF to create RDF example - Larry will provide consultation for questions - Eric will file errors found (reference to classes in patterns) - Achim will provide a high level statement on what the use of RDF is and what needs to be reviewed in the Prototype
ATTENDEES: Wendy, Dan S., Jeremy, Larry, Dan G., Jay, Flavio, Oliver, Kelly TC-138 move to Post-Prototype
TC-128 review documents to be ready for discussion The model in the prototype follows the SDA work. The model in TC-128 originated in Mpls and revised in Norway. Sampling was used tested at BAH and BLS. The starting point was translating that model into UML. The two approaches are not entirely equivilent but able to describe the same thing. Modeling view was simplified.
SampleOverview gives general approach and stages specify what is done. How are stages different from a workflow process step - probably aren't but the problem with doing that.
What is in 4 now is a process model. What Jay and Dan G are trying to do was capture the design. Is the Workflow Process a design or run-time only? We have design, algorithm and process. Design tell you the framework, the algorithm is saying these are the steps I intend to take, Process says these are the steps being taken or having taken.
Laying out the criteria and design - at the same time the process pattern is similar to what is being described here. Can the process pattern be used to describe the same thing.
The process is more imperative design is more declarative.
There isn't an exact parallel between sampling and questionnaire - In some methodology/process there is a clear definition between design and process in Sampling there is.
How does a statistician think? They don't seem to be very interested in process and looks a lot like what is presented in this material. If they then turn around and implement it they provide a bunch of attributes for them to follow.
Even if people don't think in terms of process is our process pattern generic enough to capture what they are thinking and doing. There seem to be a great number of similarities.
It might be a very high level process and may not be thought of in those terms.
It is intended to be a metadata drivable stack as input to the process to run the sample, but the direct purpose here is to document the design.
What this states are Business rules that are declaritive. So it would be interesting to see the next step.
Put it in as informative - next steps, discuss how it relates to what is currently in the model and how it fits into the broader picture. This discussion on paper.
Dan G. can write an explanation.
First of perhaps many methodologies created to define specializations of Methodology, differentiation of methodology and process and how they plug together.
Design is a class in the Methodology Pattern. Difference between "design" and design of a process.
TC-14 Documentation Looking at what we have is presented (Sphynx, examples, side papers)
TC-11 XML Clear rules of transformation from model to XML
TC-10 XMI updated July 16
TC-12 RDF? Is the actual RDF XML, how to use it, is there documentation on using it, how it fits in with binding. Eric will provide the basics. What the transformation process is. OWL, Shek work, Views. Missing overview how it fits in. Are there examples? SPAQL queries etc.
TC-13 Round-trip? Since we have an example of Australian study it would be good to look at in RDF or snipits of that.
TC-9 Model Modeling rules and how they are being realized
ATTENDEES: Wendy, Dan S., Jeremy, Oliver, Larry, Jay, Dan G., Achim, Kelly
Review examples in TC-43, TC-44 TC-44 Security and modeling difference - take a modeling approach and deal with security in the binding issue. Should document this concern. The point was not to ensure security but to recognize the issue and inform the user regarding the vulnerability of structured content. See pdf on issue Difference between structured and unstructured you are basically saying "you shouldn't use structure here" needs comprehensive enumeration Currently can validate and now you'd have to use the escape characters so another layer of translation is needed Should do it to Use different scopes of text (internal, externally intended) It would mean this feature would no longer be available Scopes are written in markdown using the div in xhtml to differentiate different kinds of texts A new attribute providing section identifier Basically uses XHTML to break into divisions and then any form of markup Add a scope attribute to both LanguageSpecificString and LanguageSpecificStructuredString enumeration and otherDefined
TC-43 Will test resolution on one usage of LanguageSpecification TC-73 Moved to Post-Prototype TC-91 will send out link to examples with questions for TC members concerning review TC-112 Update on actions to improve content TC-113 Oliver will test by running build through sphinx TC-125 Need to determine if there is anything here needing review - Wendy TC-98 Jay will follow up
Next week: TC-128 review documents to be ready for discussion TC-138 TC-14 Documentation TC-11 XML TC-10 XMI TC-12 RDF? TC-13 Round-trip? TC-9 Model
ATTENDEES: Wendy, Oliver, Larry, Jay, Kelly
TC-43 resolved Oliver will provide example prior to entry TC-44, TC-135 tentatively resolved; Oliver will provide examples TC-113 href added to images during transformation; need to verify results in read-the-docs
Sampling Methodology: Could go in as a change for the Prototype or as a supplementary document to go with review.
Still no RDF
Kelly will look over documentation to come up with a plan of what is needed from TC regarding documentation review/preparation
ATTENDEES: Wendy, Dan S., Jeremy, Oliver, Larry, Achim, Flavio
AGENDA: TC-29 Documentation/Example will be added to package TC-44, TC-135 - Oliver will look at options for next week TC-43 next week continue TC-113 Oliver and Achim will see if they can do this easily - not a show stopper, provide when available TC-112 Manual workaround for Prototype TC-98 Larry, Dan G., Jay are asked to review and rewrite - we will go with what they decide TC-17 did not get to
added issue TC-136 to capture issues to raise in review documentation
ATTENDEES: Wendy, Jon, Dan S., Jeremy, Oliver, Larry, Achim
DDILIFE-3603 Resolved DDILIFE-3604 Resolved DDI 3.3 public review We have schema in Bitbucket, new documentation in COGS, Oxygen documentation If we do the review package but the readme links to the external documentation Get the rest of the documentation into GOGS and could include the PDF version in the package YES OK Page ok and Oxygen production will be discussed after call
TC-45, TC-129, TC-95, TC-16 in review (only issue raised was with TC-45 and Oliver has resolved) TC-62 Post-prototype TC-61 not fixing TC-55 Post-prototype TC-44 oh TC-43 wlt
Next week TC-29 TC-44, TC-135 TC-43 TC-113 TC-112 TC-98 TC-17
Final sets TC-125 TC-4 TC-91 TC-73
ATTENDEES: Wendy, Dan S., Jeremy, Flavio, Larry, Johan, Guillaume
TC-95, TC-129 There is a proposal for resolution on TC-129 The proposal is correct and is in line with XKOS GSIM is in the process of redoing parts dealing with designations, codes etc. some simplifications We should review of GSIM changes across DDI when it comes out end of June ACTION: Wendy enter it
TC-92 (this was a review item and since there was no response to change I have closed it) TC-94 review and OK documentation change revised and closed
TC-16 function of variableRole a) do we need to address in prototype? b) if so, discuss We have the viewpoint but there is a role for a variableRole. Should we change the description? Change variableRole to variableFunction "Immutable characteristics of the variable such as, geographic designator, weight, temporal designation, etc." ECVE 0..n
TC-45 organization of packages related to datatypes change as noted
TC-29 structured collection of logical records make example
TC-35 relationship between ClassificationIndexEntry and ClassificationItem close and review as is in prototype
ATTENDEES: Wendy, Dan S., Jeremy, Larry, Johan, Oliver, Dan G.
Also should generate Oxygen docs to assist in how to represent in XML This is more how its modeled than how its serialized Any binding we release should have a binding specific RDF everyone seems to use the W3C documentation
STEPS: Generate the Oxygen - talk Jon, Oliver may be able to help Update with higher level content Double check readme and schemas for 3.3 dates, etc.
TC-132 - Oliver will do further checking TC-130 - closed TC-131 - appropriately classes as post-prototype TC-128 - Dan G and Jay will review TC-36 - closed TC-92 - resolved
Next week: TC-95 TC-129 Wendy will write up proposed change and post prior to next weeks meeting
ATTENDEES: Wendy, Dan S., Flavio, Larry, Oliver, Johan, Dan G.
DDILIFE-3601 - resolved, Dan S. will enter TC-105 resolved and noted added to TC-14 regarding reasoning to be included in documentation TC-130 resolved based on Wendy's drawing on TC-130. Reasoning for this level of modifications added to TC-14 for documentation
NEXT MEETING: TC-128
ATTENDEES: Wendy, Flavio, Oliver, Dan G., Jay, Johan, Kelly
Patterns: Process Pattern and Workflows: Problems with both the structure of Process Pattern and relationship to Workflows Workflow.pdf Separate the single steps from the sequence Where a workflow sequence realizes a process control step (like ifThen, Split, etc.) Remains use of workflow sequence by WorkflowProcess
We aren't doing a real good job with Temporal in terms of sequencing - we may want to look at XKOS and review how this can be used Do we add these to the Prototype or say that additional temporal work should be after the prototype Should we just be talking about synchronous or asynchronous One we include the Allan's Temporal relationships we are probably covering these - either as subclasses of ProcessControlStep or as a means of specifying the relationship within a single class structure Doesn't need to be done before Prototype but need to explain the point they would hang off of (Process Control Step) The idea of precondition/postcondition as controls - organized in a way other than IfThenElse - may be expressed temporally Ability to define both loose and tight process patterns - loose especially needed for business workflows where certain sets of things must or should be available and certain conditions for exit or next step initiation Do we need collections to define all of this or only for more formal relations
ACTION: Draft up short term (Prototype work) - decision following further review of specific changes and discussion Write up what would need to be looked at after
Collections: Problem is with realization - some of them are missing may want to add these to enumeration Add documentation of terms of relationships Codelist ends up with two different types of member and should only be codes so you have mixed member types. Contains and RelationStructure should have the same member type This is specific to CodeList so don't forbid it - review others to see if there is a problem - because of Statistical Classification etc. use as Representation - could this be better handled by an abstract Statistical Classification does not require a code Look at the further impacts - if Representation requirement then need to look if we are twisting ourselves in knots to make this work. If so we probably need to solve in Representation Is there a Statistical Classification without codes? It is optional in GSIM, but we've never seen one
ACTION: TC-129 target and source of relationStructure should be Code
DDILIFE-3519 Copenhagen mapping content Documentation: http://cdn.colectica.com/CopenhagenMapping/ Straight copy of GSIM plus additional typing such as relationship to an agent rather than just a string Publication which is an other material - should be changed to complex type with an inline and choice of OtherMaterial like earlier change for consistency Name, Label, and Description - possible future approach for simplifcation GSIM- https://statswiki.unece.org/display/gsim/3+_Object+graph Parent (there is a hierarcy in Nodes but not in classification) - should be an item in a single classification and if you have a classification used in several you should use correspondence and the map (Category rather than ClassificationItem should be shared) ADD a concept to a classification item as the classifcation item is a type of category What is the meaning of a ClassificationFamily being in the Group, StudyUnit, ResourcePackage Maybe there should be a link from within the logical product. GSIM plans to define more precisely the differences of CodeList and Statistical Classification (over next few weeks) Leave the pull request and update with those changes. Then update the pull request.
Right now choice on use of Statistical Classification is to create a separate StatisticalClassificationCodeRepresentation or to make it a choice in CodeRepresentation Wendy will make exmaples of each so people can see what the difference is.
Are we sure we have all the essential properties inherited from NodeSet and Node. Example: GSIM talks about a Part-Whole, Parent-Child Uses only those included in the Statistical Classification - by definition this may vary by level - Totality is a requirment of the statistical classification Do statistical classifications mix Whole-Part and Parent-Child at various levels within a Statistical Classification? Not within a single classification, just when there is a mix of classifications Jay will look at NAICS to verify
PROTOTYPE: Naming TC-118, TC-99, TC-84 (moved to Post-Prototype as related to TC-83), TC-63 - resolved all
ATTENDEES: Wendy, Jeremy, Dan S., Larry, Oliver, Achim
new additional labels on issues. These support an information page reflecting the progress of the TC internal review of the Prototype material: --patterns --packaging --technical --class_function (does it do what it's supposed to do) --data_capture --modification (specific requests for modification) --consistency --naming
Marketing/PAG Meetings: --Prototype goes out after end of 3.3 review period
DDILIFE-3594 closed DDILIFE-3593 closed
Issue Work: Example review TC-91 - review existing issues; note comments in TC-91; Larry will comment on specific purpose of some of the examples Technical TC-107 - resolved Role of XMI in Prototype - Achim draft up that documenation and we can then review it and ask additional questions Modification TC-124, TC-122 - resolved
DDILIFE-3519 Copenhagen mapping content
Modification TC-111, TC-23 Naming TC-118, TC-99, TC-84, TC-63 XML secondary validation Consistancy TC-119, TC-120, TC-89 Application of Design Rules
ATTENDEES: Wendy, Jeremy, Dan S., Larry, Oliver, Achim
DDILIFE-3595 language and CodeValueType - No change DDILIFE-3593 Quality Statement / Quality Standard - resolved DDILIFE-3594 [haven't entered yet] DDILIFE-3519 Classification management (Copenhagen mapping) - New metadata items are present in XML serialization but need to add in their usages - end of month Documentation: Task - rendering examples in documentation
TC-8 Please review this issue and the related documents. I am attempting to organize these into sets of related issues we can discuss and resolve during the Prototype process. Also any additional means of internal review (particularly in terms of bindings, xmi, documentation, examples, etc.). Note that some issues need some background work that perhaps smaller groups could do in preparation for a broader discussion. Others are topics where we'll want to bring in others from outside of TC. I'd like to get some sense of what we can accomplish and how the work will be distributed over the coming weeks. Need to front end issues for XMI - review and comment for discussion and decision next Thursday.
Note that MT will not be meeting for the foreseeable future so we have that timeslot Wednesday (9:00 CDT / 16:00 CET) for subgroups or additional discussion time.
NEXT WEEK: --Example review for additional issues/problems - Wendy will prepare the content for this review of updated Examples --Issues regarding XMI errors - send out email on what issues need to be looked at and commented on
ATTENDEES: Wendy, Larry, Oliver
Update on XKOS work Update on Prototype documentation DDILIFE-3593 - added comment and need a response from others not in attendence DDILIFE-3590 - approved DDILIFE-3591 - approved DDILIFE-3594 - resolved
Discussion of approach to Prototype review work: Aggrgation and Composition - issues may be too complex for prototype Reasons for doing some of these things in terms of XMI, XML, RDF, etc may be critical and we need to explore the specific reasons. Make sure and get detailed information on reasons.
Things like name changes that are better symantics for users should be reviewed in prototype Name changes to avoid conflicts with certain systems need to be changed
There should be a list of notes for people to read for review process Make a list of specific name changes that were not resolved as part of notes for reviewers
ATTENDEES: Wendy, Dan S., Jeremy, Larry, Jay, Guillaume
DDI4: (15 minutes) TC-68 resolved - fixed TC-76 resolved - no change TC-77 resolved and entered TC-81 Post-Prototype
DDILIFE (45 minutes) 3593 - go back and fix 3590 - check extension base of two items that were ext base other material 3519 - Coppenhagen Mapping - did not get to 3591 - Review in Bitbucket see commit 4ae0eda in Wendy's branch 3594 - A documentation issue. Documentation of the elements MarkedIncrement and ValueIncreatment contain their target documentation instead of the use documentation. Because both use the same base type this is confusing. Review recommended documenation change on issue and comment.
Didn't get past 3590 some work will get done remotely so we can get as much into 3.3 for NADDI as possible. TC-77 and TC-68 were entered in model post meeting.
DDILIFE-3519 Copenhagen mapping - questions noted in comments; DDILIFE-3590 OtherMaterial - management options; postpone - DDILIFE-3593 QualityStatement review and recommendation; - review not complete TC-58 (includes 38, 34, 51) - recommendation; resolved TC-70, TC-71, TC-72, TC-74 all with recommendations - all resolved TC-65 Add the 4 classes requested TC-69 determined to be post-prototype
TC-60 - to address the immediate action needed and when the larger problem of production needs to be addressed Cardinality errors with Composition have been resolved under the current framework TC-60
Aggregation cardinality should be a priority issue - review and fixed - TC-77 Example: 2nd to last comment from the bottom AggregationText Cardinality as expressed in Aggregations and Neither in Lion
In future XMI should be exported correctly - TC-78
AGENDA: Wendy, Dan s., Larry, Jay, Johan, Oliver, Kelly
OtherMaterial changes (DDILIFE-3590) - see issue Copenhagen mapping (DDILIFE-3519) DimensionType documentation (DDILIFE-3591) - OK StandardType BUG (DDILIFE-3593) - see issue 10-15 minutes on Priority Prototype issues: TC-60, TC-59, TC-57 - resolved enter set of TC-38, 45, 51, 52, 58 (dealing with reorganization of packages) - draft collective recommedation and put on next week's agenda
Timeline: End of March for final changes in 3.3; review out in April
NEXT WEEK: DDILIFE-3519 Copenhagen mapping - questions noted in comments DDILIFE-3590 OtherMaterial - management options DDILIFE-3593 QualityStatement review and recommendation TC-38, 45, 51, 52, 58 - recommendation
ATTENDEES: Wendy, Dan S, Jay, Jeremy, Oliver, Larry
DDILIFE-3589 - closed DDILIFE-3590 - resolved - Wendy will write up all changes required and post on issue for review prior to entry
Copenhagen content http://cdn.colectica.com/CopenhagenMapping/ --Where does it link into the model --How can we use StatisticalClassification directly either as a choice in CodeListRepresentation or similar to GeographicStructure/Location usage, also Catagorical --Is the target the same in terms of code? ItemCode is a property, Name, Label --From a developer perspective its not a big deal --For a large classification it keeps it in a unifying set --In using it as a representation can we make the same clarification of what is used by the representation
Do the properties make sense in terms of what has been selected, cardinality, relationships.
Do we have questions about putting this into 3.3 - general agreement it should go in --Derivation from Versionable or Maintainable - ClassificationFamily, ClassificationSeries, StatisticalClassification go in as Maintainable --FILE (DDILIFE-3592) 3.4 possiblity to collapse Identifiable/Versionable/Maintainable - file as future issue --Suggestion is to put it in LogicalProcduct and include in LogicalProductPackagingItem
--FILE (DDILIFE-3591) DimensionType documentation regarding inclusion of variables with representations other than CodeRepresentation (Geographic, Statistical, etc.)
NEXT WEEK Dan S will make a proposed list of where this would link in. Others should review and comment in JIRA issue
ATTENDEES: Wendy, Jon, Larry, Oliver, Jeremy, Dan S., Kelly, Jay
Reviewed TC issues related to Prototype identify those that were priority (could be quickly resolved and fixed) Resolved TC-33, TC-32, TC-31, TC-30, TC-28, TC-24 Those requiring changes to model will be entered by Wendy and added to the change list for Prototype internal review Those relating requiring documentation will have documentation located and relayed to RDF group.
ATTENDEES: Wendy, Dan S., Jeremy, Oliver, Larry, Jay, Guillaume, Johan, Achim, Eric, Kelly
1-RDF/OWL issues: Eric and Achim will be joining the cal
Prioritization in strategy Doing some diagnostics on model to check types were coherent Emitting OWL that used properties as defined in DDI model directly (no external namespaces) - isomorphic easy to roundtrip 3-5 figure out which foreign namespaces we want to use; write or document something that converts from XML to RDF; in practice the model always has some sort of implementation so "going through the model" may be just a conceptual approach. XMI describes the model but not an instance of metadata.
Focus should be on stategy and documentation of algorithms/approaches. Creation of tools could be part of review but not critical for prototype.
Add the strategy noted above to the strategy document and note how this relates to the Dagstuhl documents and discussions.
Protégé is one type of validation. Other forms of validation should be noted if know or a statement that we need to determine the types of validation needed.
Specific item 1: Package recognition: Useful in OWL for navigation. The use of packages and functional views could be confusing. Look at the requirement "supporting navigation". Is the package useful for this or would something else be more effective.
Specific item 2: keep contact to get on the same page, use of JIRA etc. so we can follow and comment
Specific item 3: Easier to have a simple model but this could come into conflict with how patterns are realized in the model. Can the design patterns be modeled in another way that can be recognized by the UML. Future problem.
There is the regular expression twice defined. There is a property regularExpression. 5 poly morphic properties will be filed in TC Several dialects of OWL: OWL/XML, Turtle, Manchester...etc. Which OWL sublanguage is being targeted: DL probably, might fit with RL
2-Triage on current TC issues related to Prototype Model Go through TC issues with Prototype label for next week. indicate if the issue should be addressed, documented only, or tabled for post-prototype review
3-Update on 3.3 documentation progress
NEXT WEEK: Copenhagen modeling work - where it links into the model
ATTENDING: Wendy, Jon, Oliver, Jay, Larry, Dan S., Jeremy
AGENDA: 1 - RDF/OWL Stategy for TC Review: If package is being represented in RDF this is a problem in terms of references and movement of classes between packages Would like to see early on what can be discribed for discovery purposes because members are already experienting with what they can do using the RDF This means TC needs to be involved at Step 3 4.a. What does this mean. Patterns are not currently in Bindings
This should be addressed in this strategy - doesn't look like previous work was not brought into this document Serialization in RDF must contain the same semantic content as other bindings - this is not stated explicitly here (they must be conceptually the same Round-tripping is a general requirement - requires that they be semantically identical
Don't see anything in her regarding the approach they are using regarding including other vocablaries. Mapping of generated RDF to OWL ontologies in not on this list. There were discussions of this in Dagstuhl. When do we use our own predicates and when do we replace them with our own ontologies. Part of the issue was the level of effort to do this. Part is how far to reflect RDF limitations in the model (similar to issues regarding XML).
Being able to read it into Protégé. This should be part of the strategy to verify its usability.
ACTION: Wendy will write this up and send to Kelly after review by TC by email
2-Update on 3.3 wrap-up work : Copenhagen mapping and documentation - on schedule
3-Quick review of Prototype issues filed in TC - add or refine as needed -- In terms of documentation it may allow them some additional time to do finishing/reviewed/refined -- TC's role is to determine if its fit for release -- If documentation is bad all the people reviewing will let us know
4 - TC Prototype Review: An agreement on what type of issues/bugs with DDI 4 Model should result in changes and how we should do them (as found, in sets, all at once) - we need to keep RDF group up to date on changes and also anyone doing examples - is there a way to front end this work to get any changes done asap to support other work
Changes that get made: Bugs, violations of the modeling rules Problems that may just be documented: complexity issues, multi-class relationships, improvements
Timing of changes - we should consult with Eric on this to see if he has a preference Any changes will be well documented in terms of informing RDF and example creators - Create a change log (date, change, Build number) - currently the build runs every night but only commits changes
ACTION: Wendy will write up, send to Kelly, and set up a location for a TC Review of Prototype Change Log
5 - DNS filing CV DNS filing - Strong recommendation from TC to use the DDI DNS namespace and file for an agency identification of DDI.CV. This approach becomes more important as 2 formats for publication include a DDI 3 Codelist and DDI 4 Custom Vocabulary ACTION: Wendy will write up and send to CV group, Jared, Achim and post on issue)
NOTICE: Encourage everyone to register for NADDI and send some emails on this to any groups you know of. [Wendy will send to APDU and State Data Center lists] use Jared's email
ATTENDEES: Wendy, Dan S., Jeremy, Johan, Larry
Pull request - merged
Workplan - reviewed who can do what when for next few months, adjusted for dependencies, still target of first week in April for 3.3 public review;
ACTION: post on workplan page and create issue to track assignments and progress (done)
RDF/DDI4 Model review work - JIRA tracker is being used so file issues or comment if you have a stake in this
Documentation/Marketing - just setting up and Marketing is getting involved with recruitment - more after that meeting
ATTENDEES: Wendy, Jon, Larry, Dan S., Jeremy, Johan, Guillaume, Kelly
DDI 4: We're supposed to get most of content by NADDI RDF may be delayed --binding --documentation (RDF specific) --examples --documentation of process Earlier consensus of AG: --Model itself --serializations (XMI, XML, RDF)
TC has some role for input on the serializations and can be reviewed prior to transfer of content to TC Get model serializations out for TC to look at prior to this
Criteria: --Is the XMI just out there or does it have a purpose --How is it to be used in the reveiw and can this XMI support that review - what systems are supported (ex: caviates regarding mangling of cardinality and directionality of relationships in EA) --Are these acceptable or unacceptable caviates.
XML - usages of the two sorts of XML the Library and the Views What are the relationships? Are they doing what we expect them to do - act as a profile or subset? Do the serializations to what they say on the "tin"
RDF - What is the minimum level we are willing to put out for review Valid according to the spec and opens in Protégé - needed for review This should be an iterative process - need to review They need to use JIRA so we can see their progress First call was last week, Second call today First task is to formulate invitation language to send out within the week Kelly will bring this forward to Eric
Highlevel documentation is the next essential about the relationship between the serialization and the model. What is the production system doing so that people can navigate between the RDF and the Model.
Round-tripability: Something that needs to step up a bit in the review - Something that TC should be looking at and commenting on the problems/issues Should take a look at what round-tripping means in reports from Dagstuhl 2017 Is there a clue to where the platform specific model vary from PIM in each case There is the programming that makes the binding as one point The model shouldn't be flattened for the RDF is a comment from last weeks RDF meeting. Round trip means Binding to Model to Binding
DDI Lifecycle 3.3
COGS is now taking highlevel documentation and examples It will be ready for dumping in more into it next week
(next week) 3524 - gt it in 3585 in is and needs review (in progress - hopefully next week or 2) Copenhagen mappings COGS is not yet making the high level documentation for 3.3 so that needs to be worked on -
(will start on once the near final review version is ready - not a lot of time - task that needs to be done) Do DDItoCOGS for 3.3 so we can test that out Can use the strongly typed sheet from 3.2 plus the change log to identify what changes need to be made Should Jon start uploading stuff into GITHUB or wait until Dan S. done some magic Create branch for 3.2 first
(1-2 months) High level documentation All will be done in restructured text
2018 - This year having the week long TC meeting was highly successful - bring up doing again in 2018
With controlled vocabularies have the DDI publish as DDI CodeLists - as well as current and proposed.
ATTENDEES: Wendy, Dan S., Jeremy, Larry, Jon, Oliver, Kelly, Guillaume Johan - unavailable
Update on DNS filing / DDI URN resolution Achim will be filing issues - we need to track it and nudge as needed Should CV one be revised to fit into the DDI structure (for example agency name DDI-CV)
DDILIFE-3585 see issue
DDILIFE-3524 NEW revision - Guillaume will review and comment - if OK then this is what we go with
3.3 documentation Wendy needs to do a brain dump - ASAP System is up and running after working with Jeremy Moving examples to COGS Will put up on Bitbucket to get it available for contributions Make sure people have access - they aren't beating down the doors at the moment
DID not get to. 3.3 work plan for review release 2018 priorities
ATTENDING: Wendy, Oliver, Jon, Guillaume, Jeremy, Dan S.
DDILIFE-3524 need examples see issue DDILIFE-3583 closed DDILIFE-3584 closed
Documentation: Jon needs and hour with Jeremy next week to sort out final issues. Wendy needs to do a content dump for Jon.
Short Sprint prior to NADDI. This is at the pass-off point between documentation and TC review. This is being looked at as a chance to do the following:
Wrap up any documentation issues Organize and plan work of TC review Layout DDI4 project work for the remainder of year (this would include Drupal to COGS conversion of work platform) [space has been arranged]
Oliver - remotely Jon - remotely Jeremy - maybe (Sunday is easter) focus is 3.3 almost certainly no
Announcement of sprints and ask people to participate
Should this be a TC meeting? in which case we have 3.3 work that could be done even if it is out for review. Implementation issues, software development issues.
We used to have annual face-to-face meeting and these have always been very productive we need to reinstitute support for this. If we want a TC meeting do we do this as part of a review of 4 or do we want more 3.3 line.
ATTENDEES: Wendy, Dan S., Jeremy, Oliver, Guillaume, Larry
DDILIFE-3524 Original idea was that the conditional text was interleafed with the static text. By using the current conditional text is compute text or identify text using the source parameter. But people want an IF condition that if something is true then insert this text. It's supposed to be generating the text but there is nothing there that does this. If we want both computing text and an IF statement that results in a literal text.
The conditional text type inserts a block of text How can we a conditional for literal text choices - Using the control construct within the text is overkill so we could probably make something more streamlined for this situation.
IfStatement condition and text if true - should probably go through Else to provide a default insert at minimum
Resolve to add an IfThenElse to choice in ConditionalText.
WT will draft IfElse for dynamic text.
How to indicate the position. Write documentation and examples of usage. Still want allow source parameter Want to be able to insert a combination of a text and source parameter as a result of an IF - Limit to a combination of literal text and source parameter
Reviewed DDI_L_3 entries and pulled into Master
ATTENDEES: Wendy, Oliver, Johan, Jeremy, Larry
DDILIFE 3581 - OK to remove
DDILIFE 3582 all approved with following not#4. Make sure it stays in the same order as previous when it was in in AbstraceVersionable. If not just add to each (non Abstract) to retain order
#7-9 move to reusable
DDILIFE 3524 - Check with Guilaume and others about using computation and logical expression, separating "true" and "false" responses into separate conditinal texts. How to specify which text. Definately good to be able to express specific text insert in a consistent way.