This is a report on the Metadata Interest
Group program: Converging Metadata Standards in Cultural Institutions: Apples & Oranges, which occurred Saturday June 26 2010, at 8:00 a.m. during the ALA Annual Meeting in Washington.
The three presentations provided some good insights into metadata challenges and possible courses of action. The first two dealt more closely with the theme suggested by the title of the session � aggregating heterogeneous metadata from differing sources including libraries, museums, and archives � while the third, a research report on an assessment of metadata from a cost/benefit perspective, brought in the strain of understanding user behavior and letting it guide metadata practice. This mixture came together like a good salad as evidenced by some lively question/answer at the end.
Danielle Plumer, Coordinator for the Texas Heritage Online project of the Texas State Library and Archives Commission, described the metadata education component of this cooperative digitization grant project. Over 30 institutions partnering together in 10 projects, each of which is creating at least 1,000 metadata records, were offered training at various locations in project management, legal issues, metadata standards and crosswalks (particularly content standards), controlled vocabularies, and digital preservation management. Much of the training was adapted for the audience for this project (many of whom are not librarians and who had a large learning curve) from modules developed by the Library of Congress, Cornell and others, and will be further modified to an online learning format and made available to anyone through Amigos Library Services. Danielle had some interesting comments on discoveries arising from her work with these diverse institutions: there�s nothing wrong with using MARC to describe cultural objects; LCSH is the most commonly used standard vocabulary, but is often poorly understood and some systems display it poorly; and often, metadata decisions are driven, not by the needs of a project, but by the limitations of the system/software used to create and store it.
Ching-Hsien Wang, Chief Information Officer at the Smithsonian Institution presented �Striving in Library, Archives, & Museum Converging Landscape: The power of working together�. The Smithsonian Institution encompasses 20 libraries, 19 museums, 14 archives, has recently launched a �one-stop� search center (http://collections.si.edu/search/), which will provide for the first time, the ability to search across all collections. It currently includes 4.6 million records, 445,000 images, from 40 data sources, encompassing highly diverse types of materials (e.g. books, postage stamps, audio of interviews�). In addition to simple search, the interface provides faceted browsing by object type, media, topic, name, date, place, data source, and many advanced features. �Metadata made it all happen.� They began by combining records from 8 Horizon databases, all MARC but with many differences, and as a result of that effort decided that metadata standardization needed to happen as they moved to incorporate data from the scientific and museum databases as well. The overcame challenges of defining common data elements and data typing for those elements through collaborative discussions with data providers, while respecting the diverse perspectives and traditions of different institutions. Much of the work to create unified indexes and presentation was done by programmers massaging and transforming the metadata; some MARC fields were omitted or, as with LCSH, �taken apart�. Sometimes assumed values for a particular context had to be supplied for the aggregation. They are at the end of the first phase but have much more to do; are working on iPhone and georeferencing applications and hierarchical facets, and are bringing each data source online one by one. Catalogers are learning and modifying their practices (and remedying existing metadata) as a result of seeing the outcomes.
Joyce Celeste Chapman, Library Fellow at NCSU Libraries, presented �Assessing metadata and incorporating user feedback,� a report on a research study she conducted to compare the time spent creating specific EAD elements with a study of both user behavior and opinion on the usefulness of those elements. This was a small-scale study, and Joyce was careful to state that the sample was not random and results were not generalizeable, but that the indications from them could nevertheless be useful and point to areas where a change of emphasis in metadata creation could benefit users. She also mentioned as a larger context, the recently released Final Report of the Task Force on Cost/Value Assessment of Bibliographic Control http://connect.ala.org/files/7981/costvaluetaskforcereport2010_06_18_pdf_77542.pdf . Discussion in the later question/answer period pointed out the difficulty of separating the effect of metadata practice from the effect of aspects of the discovery interface when trying to derive data on user impact. Chapman�s study was able to sidestep this problem to some degree by presenting a generic interface where each metadata element was chosen and presented separately, but she pointed out that the attempt to isolate metadata elements could result in a disjointed user experience. Timing for EAD field creation (Abstract, Bio/historical note, Scope/Content note, Subject Headings, Collection Inventory, and Other) was collected from 9 metadata creators at two institutions. A sample of end users were give 5 different tasks and their choices were analyzed; some were also interviewed on the relative importance they would place on data elements. The most striking finding was that Collection Inventory, while taking lots of time, also had high importance to users, while the Biographical note, which also was time consuming, was not ranked highly by users. There were also observations about the usefulness and duplication of information between Abstract and Scope/Content note. The user research group at NCSU is considering next steps for additional metadata assessment measures across different metadata schemes and methodologies.
In the question/answer, there were requests for sharing of metadata massaging tools/code (even if very institution-specific, people thought there�d be value in seeing how others are doing things at the code level); a general recognition that dates and subjects are among the most challenging elements to deal with in aggregating metadata; there are challenges in synchronizing / updating aggregations, especially when the source systems don�t provide timestamps; interaction between library, museum and archive folks often results in learning for all, and that all contexts could benefit from more detailed research into metadata�s value for users, however difficult that may be to assess.