Minutes from November 14, 2009

Preliminaries
Started @ 10:18 w/Syd Bauman, Kevin Hawkins (departed 11:30), Doug Knox, Michael Maben, Daniel Pitti, Chris Ruotolo, Laura Weakly, Becky Welzenbach, Rich Wisneski; James Kuhn joined ~10:30. Ended ~ 12:00.

Agenda
For details, see the full agenda.


 * Greetings and round-table introductions (10:00-10:30 am)
 * Update on Best Practices for TEI in Libraries (Syd)
 * Oral Histories: Transcribed speech v. the drama module (Laurent)
 * Generating stylesheets: Header to MARC and Header to MODS
 * Ad-hoc topics

Update on Best Practices for TEI in Libraries
KH summarized the current state of affairs leading up to the latest efforts for the group:


 * The TEI Text Encoding in Libraries Guidelines for Best Encoding Practices have been on the DLF website for years [Version 2.0 since 2005-11, version 2.1 since 2006-03. — SB]
 * these BPGs have a long pedigree, and had some financial support from the DLF (web hosting, conference calls, and several face-to-face meetings).
 * In short, this document discussed how to use TEI in Libraries. Its main contribution is its “levels of encoding”.
 * The vision is that documents may start at a lower level and progress to a higher level.
 * We are now working on version 3.0 (canonical location, current draft).
 * In this new instantiation, we have kept the concept of levels, but have abandoned the underlying theory that encoding projects will start at a lower level and go up the chain, although we still want to allow that.
 * Consequently the lower levels are no longer restricted to being a strict subset of higher levels (and they are not), although generally you add elements as you go up in levels.
 * Latest version of the document integrates the header recommendations document.

In response to a question as to whether or not we had a functional specification for interchange with other systems for the TEI Libraries BP header, KH explained that we do not have formal functional specifications, and that our main concern has been that the header is able to be generated from a MARC record. A secondary goal has been to be able to automatically generate a MARC record [for the TEI file] from the header.

We have tables mapping MARC fields to TEI elements in the spec. KH also reported that it was realized that we need tools for translation to and from MARC, and for TEI Tite to level X. KH reports that Perry Willett ran into Michael Sperberg-McQueen of Black Mesa Technologies who offered to volunteer some programming support for the group. After a brief e-mail huddle (among SB, KH, MD, and PW), he was asked if he would work on MARC ⇿ TEI header transformations.

MSMcQ graciously agreed to put in some pro bono time working on such stylesheets, a project which he has named Thutmose. Not surprisingly, our initial conversations with MSMcQ helped refine our notions of what we actually needed enormously; his consulting on the issues involved may well prove to be more valuable than the stylesheets themselves.

SB reported that the first stylesheet, basic MARC ⇾ TEI header, aka Thutmose I, level 1 has been completed and is available for review [at — SB], but as it was sent immediately before the TEI conference began, no one has looked at it yet.

This is important, as are the ODDs that SB has been creating as formal customizations for submission to the TEI Council so that the Consortium can take over maintenance of the Best Practices.

SB described the current planned architecture for the BPG document (main driver file and 4 or 5 subfile ODDs for each of levels 1-4 and maybe level 5). Further the first draft for each of the ODDs for levels 1-4 have been written. The canonical sources are available on gitHub, and the derived schemas and custom documentation (as well as copies of the sources) are temporarily available for easy download, review, and testing at.

DLF has provided support in the past with meetings, web hosting, and conference calls. DP points out OCLC might be able to give conference call support in the future. DP notes that, ala CDL and T. Catapano, some DLs use METS and TEI, and wonders if the BPG should address this. Added item to agenda for SB to report on the Optimizing Resources for Repositories and Archives effort.

The question as to whether it is best to continue work on BPGs off list, or move the work to the TEILIB-L mailing list was discussed breifly. There is concern that those not directly involved would quickly become overwhelmed, as there is a lot of detailed traffic required. It was agreed that actual work would continue off list, with frequent periodic updates sent to the TEILIB-L list.

KH reported that there had been a previous round of public comment (posted to TEI-L), from which the BPG document benefited. We briefly considered a 2nd public round of commentary. However, both KH & SB expressed a preference for the middle ground between a round of open public comment and going without public visibility at all, which is to say having a call for commentary on TEILIB-L. No one dissented, so this is our current plan. [When? — SB].

KH reports that there are two lists of outstanding issues for the BPGs: one on the wiki http://wiki.tei-c.org/index.php/Future_changes_to_Best_Practices_for_TEI_in_Libraries, primarily for issues that are pending decisions from Council. The other list of things to do is composed of “threads in my Inbox”.

SB reports that the current ODDs for BPG are finished, in the sense that they are complete and valid, but unfinished, in the sense that he is sure there are mismatches between them and at least the desired specifications, if not the currently written specifications, for levels 1–4. SB reiterated the importance of having people test the schemas very soon. level subcommittees : test ODD for your level : 2009-12-03

In response to a question by DK, KH reports that levels are more of giving people an idea of how to approach these sorts of tasks, rather than sticking [what does this mean? — SB]. SB says he phrases the same issue by noting that no one is really expected to use levels out of the box, but rather as starting point.

Harmonizing TEI Tite with the Best Practices: Is it worth pursuing?
TEI-Tite, the schema developed primarily by Perry Trollard, with input by SB, Dan O'Donnell, and Sebastian Rahtz for use in the AccessTEI program, falls between levels 3 and 4. There was significant discussion on how to incorporate Tite into the BPGs, with general agreement to make explicit that


 * 1) Use of Tite fits perfectly well within the TEI in Libraries BPGs, and
 * 2) Tite fits between levels 3 and 4.

General consensus agreed with DP that we don't want to toss control of the relationship between Tite and the Best Practices to Council.

TEI Chair Dan O'Donnell had mentioned during the TEI meeting the previous day that some changes to Tite were forthcoming to facilitate the new AccessTEI program. Since these alterations should be small and should occur reasonably soon, we agreed that we should incorporate them. [This is what my notes say, but I'm not sure what it means — SB].

DP indicated he will be raising the issue of how Tite and the BPGs relate to BOD.

Oral Histories: Transcribed speech v. the drama module (Laurent)
Both because L. Romary was not present and because we were running out of time, this agenda item was not discussed.

Linking to related, external metadata from the TEI Header
This issue has come up often. It is a difficult problem because


 * TEI Header already provides some mechanisms (like &lt;idno> or &lt;taxonomy>)
 * they vary from element to element, nothing consistent
 * differing needs: pointing to an external vocabulary, pointing to metadata itself; does external metadata supplement, replace, or duplicate that which is already in TEI Header?

There had been some recent thought that &lt;relatedItem> would do the trick, but the consensus seemed to be this would not be the case. DP notes XLink has some of what we need here; SB points out that yes, linking may be a very useful mechanism KH points out genesis of problem is desire not to duplicate metadata from main record into TEI header.

It was generally agreed that having an intra-project procedure that uses outside metadata instead of a TEI header is completely non-conformant, and outright violates Guidelines. However, since such files are not intended for interchange, it is quite reasonable to use such a procedure provided a TEI header is attached to the file before it is exported for interchange. The gist of this discussion was summed up as “you can and should do what you want in the privacy of your own repository”.

JK relates problem his project is having with this area.

SB suggests process of a sub group to hammer out at least the problem space: how much is in local header, what can or should be referred elsewhere, mechanism for reference or inclusion.

SB to put a call out to TEILIB-L for participants; BW, SB, KH, MD, and ask T. Catapano directly. SB : send e-mail to BW, KH, MD, T. Catapano, and TEILIB-L to organize sub group to work on external metadata in &lt;teiHeader> : 2009-12-15

Generating stylesheets: Header to MARC and Header to MODS
This topic was discussed only briefly as some of it had been discussed earlier (see discussions of Thutmose) and we were running short on time.

SB pointed out one major area in which stylesheet improvements will need to be made by someone (TEI-C editors, Council, this SIG) is the formatting of the custom reference documentation generated by roma or vesta. Since the stylesheets used for the web version of the Guidelines looks much better, CR and SB are to investigate using those instead or incorporating parts of them into the current stylesheets. SB : call CR re: custom doc stylesheets : 2009-11-30

Ad-hoc topics
SB introduced ORRA, the METS-iniitiaed Optimizing Resources for Repositories and Archives effort to the group. This is a yet-unfunded effort by a group including participants from the METS, TEI, DDI, and EAD communities, the goal of which is roughly analgagous to our BPG effort: to develop guidelines for using a variety of digital formats in a repository context.