Terminologies Services Meeting: Summary Report

September 12, 2007
Metropolitan Museum of Art, New York City

Goals and Objectives

The goal of the terminologies services meeting was to determine which components of a terminologies service the participants (drawn from the RLG Programs partnership and representing a wide range of different communities and roles) find the most compelling. Desired meeting outcomes included a list of the most important features and use-cases in order to help prioritize efforts of RLG Programs and OCLC Research in this area of work.

The meeting focused on the discussion of the " Terminologies Services Strawman" document. This document was distributed to participants in advance of the meeting with instruction to share and discuss it with three peers. Meeting participants reported on their conversations, and, informed by the feedback of their peers, ranked the components of a terminologies service as outlined in the strawman.

To contextualize the focus of the meeting, it may be helpful to place our effort within the larger context of Terminologies Services related issues. Some of these issues are foundational to any of the work under discussion in the strawman; some of this work may be derivative of the work on particular service components. The particular conversations we acknowledged, but ruled out-of-scope for our particular meeting:

  • Standards and best practices for technologies which allow terminologies providers to raise their terminologies to the network level.
  • Support of terminologies services for cataloguing in local tools for cataloguing, discovery and access
  • Research on how terminologies services can be exploited most efficiently to meet our user's expectations
  • Business models for sustaining the terminologies themselves and the terminologies services

Reporting Out

Each participant had the opportunity to report out, i.e. to express both the views of their institution and the views represented by their peer group. The reporting out process took most of the day, and was highly interactive. Meeting participants engaged in active give and take, with lively discussion during and between presentations (including in the hallways during breaks, and in the library during lunch!).

RLG Programs staff felt at times that we were drinking from a firehose, and our own "brief" minutes from the meeting ran to more than 12 pages. Presented here are some of the highlights from the discussion.

General observations

Access to terminologies services should be open, not just for libraries, archives and museums, but also abstracting and indexing services and anyone else. (Jenn Riley, Indiana University). Similarly, if the systems are open, people will create things we can't possibly imagine now. (Michael Winkler, University of Pennsylvania)

Raise vocabularies to the network level with URI's so we can start building services on top of them. Think about building a "thesaurus farm" (Priscilla Caplan's term). (Robin Wendler, Harvard University)

Making terms work across the LAM: 3. Services that support terminologies management and sharing could help solve pressing needs to control and coordinate terms when trying to link silos of metadata on campus. Likewise, 5. Value-added intelligence and creating relationships between terminologies coud be of particular importance to the collaborative efforts between libraries, archives and museums on a single campus / inside a single institution. (Matthew Beacom, Yale University)

People look to a handful of terminologies to suit their needs and sometimes shoehorn solutions into them, while perfectly applicable terminologies go unused. What are the possibilities for creating a registry for terminologies? (Cathryn Goodwin, Princeton University)

Image library perspective: Cataloguers spend more time on building local thesauri/authority work than on describing resources (cataloguing). Cutting and pasting from existing terminologies on the Web creates huge problems (diacritics, updating, etc.). (Elisa Lanzi, Smith College)

Observations on 1. Services that support metadata creation

Terminologies services for metadata creation should allow us to get rid of "cut and paste" and fit seamlessly into editing environments – participants felt that although there are many needs and opportunities in this area, Programs and Research would need to make serious inroads into proprietary metadata creation and editing environments. This type of effort did not seem lightweight, or achievable in the short term.

Caution, beware of speed problems: terminologies services only useful if responsive enough to be used "real time." (Robin Wendler, Harvard University)

Observations on 2. Services that support search optimization

End users should be our primary concern – if they don't find anything, they won't be back. (Amy Lucker, New York University)

Leveraging terminology for search optimization: however amended to include a multilingual approach, which could be achieved through effort in area 5 Value-added intelligence: creating relationships between terminologies. (Jonathan Franklin, National Gallery of Canada)

Creating a useful and usable name authority file is currently the holy grail in the archival community (mapping of ULAN to EAC would be of interest). Terminologies can become starting point for users: navigate from name authority to archives with desired content. (Bill Landis, Yale University).

Observations on 3. Services that support terminologies management and sharing

Problems with local terms versus terms approved via editorial process. A "local" term may be global useful, but if it's not part of an editorially controlled process, it may not be globally visible. On the other hand, there's a reluctance to share locally developed vocabularies. Observation that it's difficult to avoid developing a new vocabulary when items in a particular collection are perceived as unique.

No published terminologies are going to meet all needs. One of the reasons we have local terms is because contributing to published terminologies is so difficult. Contribution to terminologies needs to be easy. (Jenn Riley, Indiana University)

Contribution model to Getty vocabularies is a form of social network by experts—a free-for-all where contributors aren't credentialed isn't useful. (Patricia Harpring, Getty).

Similarly, it's important to allow many contributors to controlled vocabularies, yet allow evaluation of terms by keeping their provenance. (Bill Landis, Yale University)

No published terminologies are going to meet all needs. The reason we have local terms is because contributing to published terminologies is so difficult. Contribution to terminologies needs to be easy. (Jenn Riley, Indiana University)

Synchronization between terminologies "on the network level" and in local systems could prove to be a significant challenge. Example: ability to ingest entire authority records or to assign terminologies through links, then keep local term assignments appropriate and in sync when terms/concepts expand or change. (Laura Akerman, Emory University)

Observations on 4. Services that support terminologies social interactions

User tags or folksonomies may be a subversive activity to get the library out in the world. Bringing user tags in is valuable. Generally, the feeling about social tagging is that as a community, we don't have enough of it to do anything significant now, but we could have more opportunities for experimentation in the future. A lot of promise, but we're not there yet.

Social tagging is disruptive to library culture and business processes—should make us question the role for expertise and experts. How to mark authority or expertise of contributors could be an issue. (Matthew Beacom, Yale University)

Few user contributed tags implemented in image libraries, but faculty interactions and feedback would lend themselves to being captured in this way. (Elisa Lanzi, Smith College)

The magnitude of the task of describing items digitized at NARA makes user-contributed metadata an attractive option. (Kevin Devorsey, NARA)

Observations on 5. Value-added intelligence: creating relationships between terminologies

Generating terminology mappings using human intelligence, especially useful if done diligently and transparently, would be important to expose how the mappings were done. (Laura Akerman, Emory University). We have enough terms, let's start using them in a more powerful way. Mapping terms is compelling. (Amy Lucker, New York University)

If terminologies services contained "bags of words," electronic tagging of born-digital material might be possible. Getting humans out of the descriptive process as much as possible is an important goal. (Bill Landis, Yale University)

Meeting Outcomes

During the reporting out process, there was a lot of give and take, and it was clear during reporting, during discussion, and over lunch that participants had been swayed by their colleagues' opinions in many ways. Additionally, several institutions had given two first place votes, lumped votes under work areas instead of work items, which made tallying the initial votes difficult. Therefore, RLG Programs took a final vote on the Strawman proposals. Participants were given two votes, the first worth two points, the second worth one point.

And the winners were:

2.b Leveraging terminology for search optimization (25 votes)
3.c Combining local, shared or published terminologies (7 votes)
5.b Generating terminology mappings using human intelligence (7 votes)
5.a Generating terminology mappings using data mining techniques (4 votes)
1.a Browsing, searching and retrieving terms (1 vote)
1.b Displaying part of an authority record (1 vote)

Wrap up Discussion

RLG Programs will consult with colleagues in OCLC Research in order to make a doable plan to go forward. This will involve developing one or two prototypes, and testing with a subset of meeting participants. RLG Programs will also produce a summary report from the meeting.

Meeting Participants:
Laura Akerman (Robert W. Woodruff Library, Emory University)
Matthew Beacom (Yale University Library)
Kevin DeVorsey (NARA)
Mary Elings (Bancroft Library, UC Berkeley)
Jonathan Franklin (Library & Archives, National Gallery of Canada)
Jeremy Frumkin (Oregon State University Libraries)
Cathryn Goodwin (Princeton University Art Museum),
Patricia Harpring (Getty Research Institute)
Andrew Houghton (OCLC Research)
Bill Landis (Yale University Library)
Elisa Lanzi (Imaging Center, Smith College)
Amy Lucker (Institute of Fine Arts Library, New York University)
Merrilee Proffitt (RLG Programs, OCLC)
Jenn Riley (Digital Library Program, Indiana University)
Daniel Starr (Thomas J. Watson Library, Metropolitan Museum of Art)
Günter Waibel (RLG Programs, OCLC)
Robin Wendler (Harvard University Library)
Michael Winkler (University Library, University of Pennsylvania)

Meeting participants discussed the Strawman with the following peers:

Murtha Baca, Getty Research Institute
Penny Baker, Sterling and Francine Clark Art Institute
Jennifer Bowen, University of Rochester
Randal Brandt, University of California, Berkeley
John Butler, University of Minnesota
Priscilla Caplan, University of Florida
Anne Champagne, Art Institute of Chicago
Joan Cobb, Getty Research Institute
Erin Coburn, J. Paul Getty Museum
John W. Chapman, University of Minnesota
Karen Coyle, Consultant
Ross Day, Metropolitan Museum of Art
Jolene M. de Verges, Massachusetts Institute of Technology
Toby Dodds, Smithsonian Global Sound
Deirdre Donohue, The International Center for Photography
Christine Droll, Nelson Atkins Museum of Art
Jocelyn Gibbs, Getty Research Institute
Ken Hamma, J. Paul Getty Trust
John Hansen, American Museum of Natural History
Stephen Hearn, University of Minnesota
Diane Hillmann, Cornell University
Nancy Hoebelheinrich, Stanford University
Lynn Holdzkom, University of North Carolina at Chapel Hill
Arwen Hutt, University of California, San Diego
Marty Kurth, Cornell University
Eric Lease Morgan, Notre Dame University
Michelle Light, University of California, Irvine
Donna McCrea, University of Montana
Irene McMorland, Federal Reserve Bank of Atlanta
Jessica Milby, Philadelphia Museum of Art
Ron Murray, Library of Congress
Terry Reese, Oregon State University
David Reynolds, Johns Hopkins University
Tricia Rose, University of California, San Diego
Will Real, Carnegie Museum of Art
Joe Shubitowski, Getty Research Institute
Kelcy Shepherd, University of Amherst
Ryan Scherle, Duke University
Sarah Shreeves, University of Illinois, Urbana-Champaign
Melanie Wacker, Columbia University
Karen Weiss, Smithsonian Archives of American Art
Layna White, San Francisco Museum of Modern Art
Ann Whiteside, Massachusetts Institute of Technology