September 12, 2008
By Eberhart, George
From folksonomies to federated searches, reference databases are constantly evolving. Meet seven industry leaders who are driving those changes. Seven leading publishers shared their insights on the future of reference databases at American Libraries' second annual "Speaking Technically" panel at ALA Annual Conference in Anaheim. Moderated by American Libraries Direct Editor George Eberhart, the panelists talked about their new products and ideas for enhanced services. On hand were (left to right, above) JIM DRAPER, vice president and publisher of Gale/Cengage Learning; STEPHEN RHINDTUTT, president of Alexander Street Press; CHRISTOPHER WARNOCK, CEO of Ebrary; MICHAEL GORRELL, chief information officer of EBSCO; KEVIN OHE, editorial director for electronic products at Greenwood Publishing Group; ANTHONY PAREDES, vice president of Capital IQ, a division of Standard and Poor's; and LINDA GOLDBERG, vice president for user experience at ProQuest. Watch a video of this program at AL Focus, at alfocus.ala.org.
LINDA GOLDBERG: I want to see Dialog as a service that makes users fascinated, not only with the tool that they're using for search and retrieval, but with the information that they find there . That is the goal-making the experience of retrieving, locating, and identifying information fascinating, inspiring, and engaging.
How much time do you all spend looking into new technologies for information delivery?
MICHAEL GORRELL: EBSCO is constantly looking at new technologies and new user experiences. We just released EBSCOhost 2.0, the first major revision of our user interface since 2002. In the past sixyears, we've also added many products and features based on our research.
Do you use any focus groups?
GORRELL: We use many focus groups, usability studies, and user testing, all the way from guerrilla user testingpulling people off the street or off the campus-to structured laboratory testing. The Kent State University School of Library and Information Science has a Usability Lab with Tobii eye-tracking hardware, and we did some formal studies there.
JIM DRAPER: We do that all the time at Gale. About two years ago we conducted an 18-month study of how graduate students and undergraduates use and interact with literature products, and we learned some fascinating things about how they bounce around within products-what they call "flicking." As we watched their hands and eyes interact with the screen, we realized that we could do a much better job for the end-user experience.
What are the best ways for your customers to tell you about enhancements that they'd like to see?
CHRISTOPHER WARNOCK: In every customer visit that Ebrary makes, we ask for feedback. We have a trial program that we put potential customers through. During this program we ask for feedback on the system and how they see it. Librarians are forthcoming about what works and what doesn't work. We also conduct surveys with librarians, faculty, and students.
ANTHONY PAREDES: We do the same at Capital IQ and Standard and Poor's. For the last eight years, we've had an option for customers to submit feedback or request data and usability enhancements directly on the platform. About every two months, we add new functionality into our products, all driven from client feedback.
Gale has just introduced a federated search service that allows users to search for articles across your many databases. How does that work?
DRAPER: OneSearch is a complement to our PowerSearch program that has been in production for about two years. It's a high-end federated search that looks at metadata across all of the Gale databases. Librarians can sign up for the PowerSearch Plus and look in other vendors' databases. For example, ProQuest products can be viewed through our interface, and it harvests their metadata, making for a much more powerful search engine. The inspiration forthat came from our customers.
What other search technologies will you be implementing over the next few years?
STEPHEN RHIND-TUTT: Video is extremely cumbersome to search; although Google has made some inroads, the proportion of video they've put up is relatively small and in the public domain. When you go into a bookshop, you can find both a public-domain $10 version of a classic novel by Charles Dickens as well as the $45 Oxford University Press scholarly edition. At Alexander Street Press, we think there's a similar opportunity to create what we call a "critical video edition." It should serve as a research tool, not just by Alexander Street users, but by others through OpenURL and other protocols.
Going forward, perhaps we'll have 200 different search mechanisms to interact with everyone else's database.
How will metadata fit into this new federated search concept?
WARNOCK: Metadata is a funny thing, because in some cases it costs more than the information it describes. Researchers are studying ways for full text to be used to either eliminate the need for metadata or to allow new forms of metadata. There are hundreds of ways to go inside a document, extract information from the full text directly, and present it in a new interface.
Another way to do that is to add user-sourced metadata. Will folksonomies work as a supplement to structured thesauri?
RHIND-TUTT: Folksonomies are interesting, but too often they are given more credit than they deserve. They are useful in such applications as del.icio.us, Connotea, and CiteULike, but they fall prey to a recurring historical cycle. Initially, e-mail was shared effectively by a very small cadre of people; now, 90% of it is spam. In late 2005, Alexander Street Press launched a folksonomically oriented database on women and social movements. Rut we found that when the size of the user community was only 500 or so academics, folksonomies were not that useful except as adjuncts to an existing taxonomy, or as a help for keyword or full-text search. They are not a silver bullet.
Can the library profession help in sorting some of this out?
RHIND-TUTT: Indeed. I remember when del.icio.us began, most of the tags you saw were good because the early adopters were librarians and others who were very selective. But increasingly, less scrupulous people have put, in effect, spam tags into del.icio.us. They've added some "recommended tags," but you have to ask yourself whether that in effect is doing rudimentary cataloging or creating a taxonomy in the classical sense. The Google Image Labeler game used to be fun when it started in 2006. It assigns you a partner, shows you a set of images, and you try to guess what the labels are in order to improve Google image tagging. I played it again about two weeks ago, and twice the partner I had was simply giving me spam responses, whatever the image.
DRAPER: I'm convinced that folksonomy needs to live as a complement to the main taxonomy, and then we can use that to help create effective finding aids.
WARNOCK: When telephone switchboards first came out, companies quickly learned that teenage boys could not be telephone operators because they were all too happy to make connections incorrectly, randomly disconnect people, or listen in. When we were talking with librarians about introducing folksonomies into Ebrary, we had a number of people point out that competitive students might benefit from mistagging or hiding information, just as students sometimes mis-shelve books so that they have exclusive access to them.
GORRELL: The vast majority of our users don't use our services because they love them, they use them because they have to. I just can't see college students tagging articles inside EBSCOhost or ProQuest.
GOLDBERG: ProQuest is looking at tagging as a way to allow users to supply content, but only in a way that we can allow an invited community of people to supply tags.
GORRELL: Even if we identify a cool, new, or buzzwordy technology, we don't necessarily need to take advantage of it. Amazon.com has a great user experience, but not all the things that Amazon does would apply to what EBSCOhost does. You can't jump in because the technology seems to make sense in a certain area. It has to apply in context.
What about blogs as add-ons?
KEVIN OHE: At Greenwood Publishingwe started a blog with a new product, Pop Culture Universe, which is basically a reference-book database. Recause it is book-based, we can tell librarians and students that it is a legitimate place to go to get information that "counts." We launched the blog for two reasons: We can develop an active user community around Pop Culture Universe, and it enables us to keep current in a way that a book cannot. Our bloggers can write about new pop-culture topics instantly and refer users back to our site, where they can get the good content.
DRAPER: At Gale we launched a reader's advisory database called Rooks and Authors. Users can open an account (it takes about two seconds) and write their own book reviews in their own My Reading Room space . We were hoping a few people would write reviews. Well, we had thousands of them coming in every week, even though the product was new. That clearly demonstrates that in the sphere of books and literature, our users like to share. I don't think I could do the same thing in a history database quite as well, and I certainly couldn't in science. RHIND-TUTT: In our video and music collections, we've made it possible for users to place bookmarks within a video or audio performance, string them together, and annotate them. Currently we have more than 25,000 academics who have created miniature course packs that they can distribute. It allows students more convenient access to media that historically are hard to dig into.
Capital IQ has something called "Six Degrees of Separation" tools. Is that something that might be applicable to other databases?
PAREDES: The platform was started by investment bankers who were looking to find deals and use contacts that they might have had at a previous company. It allows you to locate your own business relationships and contacts. The system will map out those relationships for you. You can ask it to look for, say, an oncology specialist who resides in or works for a company located within a specific area and has certain types of credentials. The information comes from a variety of sources, including researchers who compile data the old fashioned way, then categorize it and smart-tag it.
What about translation features?
WARNOCK: Machine translation is not yet at the point where it can take a phrase in one language, translate it to another, and translate it back the same way. Usage studies show that as soon as you encounter a word or phrase that you don't understand, all of your reading comprehension from that point forward is compromised. Ebrary does offer this in order to help people clarify concepts or context.
DRAPER: Gale has machine language translation in nearly all of our products now, and that's a requirement from many of our customers. We've incorporated Mandarin Chinese, Korean, and some of the traditional European languages, and we're building this out all the time. It's an absolute must in the next year or two, if you really want your products to get wide distribution.
WARNOCK: Publishers are in a position to do that for their own materials, but Ebrary has specifically avoided full-document translations in case it impedes on rights that have been sold to other countries, or on the interpretation of the author's work.
GORRELL: EBSCOhost does have full-text machine language translation. We give a "your mileage may vary" message as you click the translation, because it's not going to be exact. We don't want people doing surgery based on a poor translation.
Is there anything that you've got lined up that will help librarians better promote the value of your resources?
GOLDBERG: We are taking a look at the spectrum of content now under the ProQuest umbrella, given the 2007 merger of ProQuest and CSA. We need to present these products in a way that makes it easier for librarians to select the products that are best for their users. ProQuest also supplies downloadable marketing kits that help many types of libraries-corporate, academic, public, military-connect with their user communities and raise their awareness of the library's resources.
OHE: Greenwood has developed multimedia "research tutors" that help students find and evaluate content. Librarians often don't have time to spend teaching this to students, teachers assume the librarians are doing it, and students wind up with no concept of how to do effective research. The research tutor is a quick link that anyone canuse. Students can see a slide show with a voiceover that walks them through a topic such as plagiarism. We also have "wizards" that step kids through the writing process.
Does Ebrary have a new Java reader coming out?
WARNOCK: We gave our first full demonstration publicly this morning. It's not just a new Java reader; it's part of a new suite of technologies we call ISIS (Integrated, Scalable Information System) . This adds chapter-level or subchapter-level search results instead of just document-level results, and it's more ADA- compliant. ISIS has been in development for four years, and it will all be live before the end of summer.
How are database users changing over the years?
DRAPER: I read a fascinating 2008 report, Information Behaviour of the Researcher of the Future, by the University College London's CIBER group, showing that students-high school or younger-who intend to do a research project on, say, George Washington will begin by performing an image search. Think about that for a moment. When I was their age, it would never have crossed my mind to need a picture of George Washington before knowing when he was born or whether he cut down that cherry tree. It indicates an absolutely fascinating shift coming in the way in which people perceive information. All of us need to begin to grapple with this.
WARNOCK: Here is a counter-perspective: It's the lack of a cover and frontispiece. In our generation, our only option was to pick up a book. You'd und a book on George Washington through the card catalog, pull the book off the shelf, and nine times out of 10 you'd find a picture of George Washington on the cover or on the frontispiece. Thirty percent of our brain goes to processing images. I think it's just lack of a format.
Audience question: I can remember in the days when I was working the reference desk and a young person came to the desk and asked me for a photograph of George Washington. That is what we now call a "teachable moment." When do you think we can make those teachable moments happen in reference databases so a young person can learn that photography didn't exist at the time?
DRAPER: I can imagine an "in your face" approach where if you search for a video of Abraham Lincoln you would learn during the search process that video did not exist in the age of Abraham Lincoln. But saying that politely might be a challenge.
"Folksonomies are interesting, but too often they are given more credit than they deserve."
"Students-high school or younger-who intend to do a research project on, say, George Washington will begin by performing an image search. Think about that for a moment."
Copyright American Library Association Aug 2008
(c) 2008 American Libraries. Provided by ProQuest LLC. All rights Reserved.
On the Net: