Notes from OPAC 2.0 Conference

Archived Presentations:

Presenters

• Michael Kaplan (Primo)
• Gregory Crane (Perseus)
• Laurie Allen (PennTags)
• Abby Blachly (LibraryThing)

OPAC 2.0 and Beyond: Can Librarians Succeed as Counter-Counter-Revolutionaries

Presenter: Michael Kaplan, Digital Products Technical Support Director at Ex Libris. Prior to 2000, Associate Dean of libraries and Director of Technical Services at Indiana University, Bloomington. At one point served as chair of PCC and Big Heads.

Today mostly trying to sell us on Primo, i.e., Ex Libris's answer to MediaLab's AquaBrowser and Endeca's Information Access Platform . Application of Gresham's Law to cataloging: Cheaper or quicker will generally prevail over more expensive or slower. There is a Kulturkampf raging between administrators and Technical Services staff, as evidenced in Karen Calhoun's position papers, LC's series tracing decision, University of California's "Rethinking how we Provide Bibliographic Services ...", OCLC's "Perceptions of Libraries and Information resources", animated discussion on the NGC4LIB discussion group, and Jeffrey Beall's recent comment in LJ ("Stop the war on metadata"). Kaplan believes that the fight is over. Since at least 2001, the demand for human-generated metadata has declined while "full-text is king".

How should we respond, when events always seem to overtake the profession. Vendor community is too small to compete with Google, and implement new technologies in a timely manner. Some bright spots though. Major 2006 milestone was OCLC exposing library data on Web through OpenWorldCat.org. Presentation and social networking functions also getting better.

Open URL link resolvers such as SFX, federated searching (e.g., MetaLib), digital asset management systems (e.g., Digitool), Electronic Resource Management systems (e.g., Verde), are growing in importance. ." Cites "The Future of Integrated library Systems" by Marshall Breeding, who asserted that the ILS, after having matured over 2 decades, reached a crisis around 2000, as it remained overly fixated on print resources. Also, as Roy Tennant has put it, "Librarians like to Search, but users like to find."

The most important feature for a 2.0 OPAC is relevancy ranking, Kaplan suggests. Other key elements include faceting, clustering, FRBR display, social tagging, and recommendations. In order to optimize these features, the user experience must be "decoupled" from back-office functions, i.e., data creation and maintenance from discovery (slide 62).

My question: You began by suggesting that the library community can never catch up with Google, but then went on to say we should purchase Primo or something similar in order to stay competitive. Aren't you contradicting yourself? Wouldn't it make more sense simply to incorporate Google applications where ever they outperform library-specific ones? At the same time, shouldn't we participate in open source software movement in order to supplement the small number of developers employed by library system vendors? Also, shouldn't libraries be focusing their energies where we really have a competetive edge, namely, educational mission, access to authoritative information, commitment to intellectual freedom, and market ourselves accordingly? Despite what it's literature says, Google's mission is to turn a profit for shareholders (witness their willingness to censor information in China). This doesn't make Google evil, but it should remind us that we occupy a very different niche in the information economy. Kaplan responded that "libraries have the moral high ground; Google has the technological high ground"; and libraries need to be credible in both areas. So, I'm not sure he understood my question. To be fair, though, since he represents a for-profit vendor, he may not be able to identify himself fully with the non-profit, educational perspective I was suggesting.

The Library's New Edge: Crossing the OPAC's Threshold

Presenter: John Blyberg, Systems Administrator and Lead Developer for Ann Arbor District Library (AADL) . Recognized by Library Journal as 2006 "Mover & Shaker".

Current library OPAC is "Web 1.0 in every way", i.e., not very customizable, attractive, extensible, or accommodating of user-driven improvements. Feels the same way about library Web sites. Wants people to remember that the OPAC and Web site are the library's public face.

The new OPAC needs to meet people where they are, rather than expecting them to come to us. Example: virtual catalog cards that allow readers to write marginalia, and deposit personalized cards in their personal collection. Don't underestimate the ingenuity of patrons. Give them more control over library tools, and see how they develop them. Then incorporate the best of these new developments into general system.

Recommends Hinchcliffe's (web2.wsj2.com) 10 ways to take advantage of Web 2.0: (1) encourage social contributions with individual benefit; (2) make content editable whenever possible; (3) encourage unintended uses; (4) provide continuous, interactive user experiences; (5) offer content as feeds and/or web services; (6) let users establish and build on their reputations (best example: slashdot.org); (7) allow low-friction enrichment of information; (8) give users right to remix; (9) reuse other services aggressively; (10) build small pieces loosely joined (cf: David Weinberg book title).

Fundamentals of Library 2.0 Web design include: (1) single sign-on; (2) open standards (vendors need to move more quickly on this); (3) OSS (philosophically the right thing to do, i.e., managing access to our own data); (4) integrated OPAC; and (5) social software.

Suggestion for getting "beyond the website" include: (1) designing for innovation, e.g., by using and releasing APIs, encouraging community development and lowering barriers to user and staff creativity, reaching out to potential partners such as local library schools, and education; (2) recognizing importance of "gadgets" in users' lives, including IPv6, cell phones, PDAs, PMPs, PGDs, smart homes, smart cars, and remembering that "knowing technology means knowing how to take advantage of it"; (3) incorporating services such as Google books/maps/gadgets, Amazon, Greasemonkey, etc. (Blyberg used Google Gadgets API, for example, to sort titles by popularity, etc., and mash-ups that combine AADL records with Google Book images). Byberg disputes that we're in competition with Google, but maintains that even if we were, we have competitive edge insofar as we really know our communities.

What Happens When the Books Start to Talk to Each Other?

Presenter: Gregory Crane, editor in chief of the Perseus Digital Library at Tufts, currently administering $2.7 million grant from Digital Library Initiative. According to the bio sheet, "His current research focuses on 'computational humanities' and how this new field can help to democratize information without compromising intellectual rigor."

Flattered audience by saying that "the one thing that librarians do really well is cataloging," and is reluctant to meddle in this activity, but "cataloging data is essential" to what he wants to accomplish.

Idea of "incunabula" is that a new publishing medium starts off by imitating the old, e.g., the printed book at first mimicked the manuscript codex. (Interestingly, the incunabulum Gutenberg Bible featured moveable type but also hand-crafted illuminations that subverted the productivity gains of the new technology.) Today, a good example of incunabula is the PDF file which is designed to look (and largely function) like a printed book. The post-incunabula Web, however, tries to separate data from presentation. It supports recombinant dynamic data elements, and prefers open-source over locked-down code, and features books that can "talk to one another".

Recombinant data requires a substratum of structured metadata. Crane's work requires use of metadata not normally found in library catalog records, e.g., native citation scheme such as chapter numbers. Perhaps catalogers should be in the business of providing such detail, since they prove especially useful in systems interoperability. In the humanities, native citation schemes are extremely important for collocating various editions along with secondary sourcess.

Regarding "dynamic data", Crane used to disparage Wikipedia, but now marvels at its capacity for self-policing and correction. Though professional academics are traditionally skeptical about this kind of thing, amateur historians and genealogists turn out to be exactly the kinds of people you want scrutinizing your data (e.g., disambiguating place names identified through automated TGN matching--see below).

The idea of books talking to one another was suggested in the artificial intelligence work of Marvin Minsky. What Crane is discussing is less futuristic but extremely useful. He demonstrated for example, a context-sensitive dictionary of morphological anaylsis that pops up while reading a Greek text in the Perseus database.

Crane stressed the importance of "named entity analysis" whereby a block of text is auto-tagged, and place names mapped against Getty's Thesaurus of Geographic Names (TGN) and Google Maps. Translations can be done on the fly through automated extraction and pairing of blocks of original text with blocks of translation text.

Conclusions: librarians are central to these new developments, and opportunities to contribute are vast. There is an especially great need for authority lists and other forms of disambiguation. He mentioned CIDOC-CRM (International Council of Museums International Committee for Documentation Conceptual Reference Model), FRBR, MODS as key resources. Sustainable cataloging in era of scarce labor will require Wikipedia-like collaboration plus automated analysis. Let machines to most of the analysis and mapping, and let human experts set the algorithms and tweak the results. The larger community can also contribute their interests and experience to building up the knowledge base.

Penntags: Building a Social Tagging System in an Academic Library

Presenter: Laurie Allen, chair of PennTags Development Team (1 librarian, 1 programmer) and Social Sciences Data Services Librarian at UPenn's Van Pelt library.

PennTags employs Oracle, bookmarklets, javascript, "triplets" (RDF?), HTML & RSS output engines, and AJAX laid on top of catalog results. Every PennTag page is also an RSS feed.

Users are able to assign tags, find resources using tags, write long annotations, share links, annotations, and tags, and have annotations displayed within OPAC and PennTags pages. Students use PennTags create simple bookmarks and annotated bibliographies as class assignments, librarians to prepare research guides, while enthusiasts create "shadow catalogs", such as the 700 items identified as relating to 20th century classical music in cinema.

PennTags has not officially been released to the public, and Allen looks forward to an upcoming marketing campaign within university, collaboration with other universities, and development of an instrument to measure influence. Future releases to include DOI support, image functions, output integration to other library resources, export to delicious, Endnote, rtf, Refworks, Bibtext, etc., an open access API, export option in RDF, and better data mining capability.

I asked about copyright protection of licensed resources that were tagged and then exported. Allen answered that only the tags and URLs get exported. Anyone trying to open a linked resource would still need to be authenticate as a licensed user.

Could something similar be set up at Yale using Voyager and Sakai?

The Future of the Thing: LibraryThing and Social Cataloging

Presenter: Abby Blachly, head (and only) librarian at librarything.com.

Tim O'Reily describes Web 2.0 as harnessing the collective wisdom and activity of the masses. Library Thing (LT) applies this principle to the public's love of reading, and to the the networked libraries and bookstores that support it. LT's simple search box points to Amazon, but Z39.50 allows searching across multiple library catalogs as well. The database includes 7.2 million user-tagged catalog records (of which many are duplicates) and 87,000 book reviews. By contrast, she suggests, tagging on Amazon has been anaemic, because it's like "volunteering to fluff pillows at the local Sheraton". With LT, users get to shape and personalize their own catalogs, and, because the API is made available, they can create their own derivative applications. Another important part of its success has been a low barrier to participation. The single requirement for a free account is establishing a unique username and password.

Interesting features include the identification of prospective "soul mates", i.e., members determined through machine matching to have book collections most similar to your own; a portrait gallery composed of user-submitted author images; an interface translated by volunteers into 30 languages, and a communal form of authority control whereby users submit commands to "Combine" and "Separate" access points.

Why tags rather than LC subject headings? A year after event, "Hurricane Katrina" still not in LCSH [is that true?]. By contrast, an LT tag cloud begins forming immediately. This isn't to say that LCSH is unimportant, and LT in fact supports it.

At least one cultural heritage institution has adopted LT as its official OPAC: New York's Museum of Comic Art uses it to display its 1,040-title book collection.

Given that most users have free accounts (i.e., the version usable for up to 200 titles), I asked about cost-recovery. Abby explained that in addition to the membership fees from heavy users, they have recently received venture capital.

Currently the system is optimized for books, but work is underway to support other media more effectively.

Given access to API, could certain features be lifted out of the full program and incorporated into, say, a Voyager environment? E.g., potential use as recommendation engine?

dsl 11/06