"Special librarians" and enterprise knowledge management

Emedia Professional, April, 1998 by Robert J. Boeri, Martin Hensel

When electronic media emerged in the late 1970s, so-called "special librarians" began to provide specific services to commercial enterprises with growing information management needs. Such librarians developed categories for research documents, designed keywords to help engineers find specific documents, and performed information searches on the few online services then available. Content-based retrieval as a common component of enterprise document management, and the Internet, would not be routinely available for another 20 years. The services special librarians provided then remain highly relevant to today's enterprise knowledge management efforts.

According to the Special Library Association (SLA), special librarians "are information resource experts dedicated to putting knowledge to work." In addition to performing expected search services for their patrons, the SLA includes two other services that can dovetail with corporate knowledge management projects: "evaluating and comparing information software and sources of data prior to purchase" and "creating databases for organizations to access their internal information." These two librarian services can prove key to knowledge management projects, especially those including selection and deployment of search systems.

What kinds of activities in a cross-discipline business team--including, but not limited to, information technology professionals--do search vendors themselves suggest as roles needing special librarian skills? Text retrieval is a cornerstone technology for knowledge management, so the suggestions of two mainstream text retrieval vendors, Verity and Excalibur, should provide useful answers.

WHERE SPECIAL LIBRARIANS AIR THEIR RETRIEVAL WARES

Excalibur Technologies recently inked a pact with Electronic Online Systems International to use Excalibur's RetrievalWare search software in EOS' Q Series library information product. The Q series will be able to sort through hundreds of thousands of book and journal entries held in a library's automated "card catalog," using natural language queries. "Searching an automated library, such as the type that EOS creates, is a natural application for our technology," says Patrick Condo, CEO and President of Excalibur Technologies. No matter how natural the query language, or how automatic the search system, librarians will log many hours preparing electronic documents for electronic searching and helping patrons focus their searches.

Implementing Verity's search system in a corporate application makes it clear that librarian skills are essential to performing such a task effectively. Although you can use Verity's search systems pretty much "out of the box," customizing the system provides very high levels of recall (finding all the relevant documents the query asks for) and precision (avoiding false hits). Verity's system uses the notion of concept searches (called "topics"), which are hierarchical queries combining Boolean search operators and some of Verity's own operators. The power of topics is that you can combine them, like LEGO[R] blocks, in increasingly complex concept structures. Managing libraries of topics also requires experience in cataloging information.

More basic even than concept structures are fundamental questions of language, meaning, and character sets themselves. Even if you have the luxury of managing textual information in only one language--an ever-rarer option--you probably will need to consider and customize these features of search systems. Does your corporation use a specialized vocabulary, as most do? Creating a customized thesaurus will make searching more productive. What is the "lexical character set," those characters that you can search for and those that you want specifically to exclude? For example, if you want routinely to search for copyright symbols, you may want to search for the pie-font character "[C]". Will users perform case-sensitive searches, such as for customer names like "McDonnel"? Such searches may increase index sizes and have other side effects. Do you want to search for numbers? Can numbered searches include non-numeric items, as with designations in a part catalog that combine numbers, letters, and other characters, such as dashes? And will users want to combine free-form text searches with structured attribute searches (such as author's name, document title, and creation date)? A cross-functional team without a librarian would be hard-pressed to answer such questions correctly--if they think of them at all.

A VOICE FROM THE TRENCHES

Experian Corporation, formerly a TRW credit reporting company, provides credit reporting services for consumers, business, and real estate. Joyce Adams, a special librarian for 20 years, is the company's Web site operations manager. Adams says that special librarian organizational skills helped her organize the Web site as it increased in size from a token site to over 160 pages, and continues to grow. "One of the first things I did in January was to register our Web site with the search engines," Adams says. "Who better to index page content than someone who has spent years searching for page content? Although the actual process of registering a page is not difficult, the challenge came in using the right descriptors to characterize our site, depending on the focus and methodology of the search engine."