BackTalk: Stop the War on Metadata
By Jeffrey Beall -- Library Journal, 7/15/2006
For decades, metadata has been at the core of librarianship. In recent years, however, there has been a growing revolt that threatens metadata-enabled searches in favor of full-text searching. If the revolt succeeds, full-text searching will make metadata as rare as card catalogs, thus leaving information discovery at the mercy of keyword searching and whatever murky relevancy ranking a search engine provides.
Without question, full-text searching has its advantages. Still, we should not be so quick to abandon metadata. While full-text searching enables quick searching of millions of documents, it also suffers from inherent flaws, such as low precision and the tendency simply to miss a lot of relevant information—very often the exact piece of information a user may find most useful.
Battle plan
The theory goes that as more and more information becomes available online, information seekers will find everything they want using full-text search engines, thus making metadata redundant and unnecessary. Already the simple search box has become the standard, and metadata-enabled search platforms that provide for more precise searching, such as by author, title, or subject across multiple authoritative databases, are at risk. For example, citing full-text searching, managers at the Library of Congress (LC) recently announced a controversial new policy that would eliminate the ability to search by series title in LC’s online catalog.Indeed, when Google states it mission is to “organize all the world’s information” through this simple search box, it reveals its battle plan: Google wants everyone to believe that its simple search box is all that’s needed for efficient and precise searching. It’s been remarkably effective in promoting this idea so far. The simple search box now enjoys the popular perception that it is an efficient means of finding information.For libraries to rely on keyword searching, however, represents the institutionalization of serendipity as a search strategy. It is the electronic equivalent of browsing the stacks without any plan and hoping to find needed information. The simple search box is deceptively attractive. It often seems to fulfill one’s searching needs. But in practice, if you don’t know a valuable resource exists, chances are you’ll never know it if your search doesn’t return it—and prominently—in your results.
Institutionalizing imprecision
When we evaluate good searching, precision—that is the ratio of relevant hits in a search to the total number found—is key. The next time you do a search on Google, see how many millions of hits you generate and you’ll observe poor precision firsthand. One of the reasons for poor precision in full-text searching is its reliance on language. A single term can mean many different things. For example, a person interested in weaving might type the word looms, only to find many results that have nothing to do with weaving but using the word looms in a different context. This is called a problem of polysemy.
This reliance on language is a major weakness of full-text searching. There is also the problem of synonymy. For example, if a searcher needs information about plant science, but the best resources call it botany, then the searcher will likely be unsuccessful in his search. Our language is rich, and we often use many precise terms to represent a single concept. Full-text searching, however, is inherently imprecise in its execution.
Good metadata, on the other hand, adds value to information. It enables precise searching by author, title, and subject, and combined with a good search platform, such as an online catalog, it can neatly colocate works so that they display on a results screen together with other similar or related works. The business world understands the value of metadata for its information and is investing heavily in it. Libraries, as custodians of information, must do the same.
Defend metadata
Ironically, many of the top generals in the war against metadata are librarians and information managers themselves. These managers are making drastic cutbacks in personnel dedicated to metadata creation. They have created databases and repositories that rely on full-text searching for access to their content. Yet while they’ve proven themselves “efficient” and “fiscally tight” project managers by creating such search-based systems, the reality is that such systems strive to be only good enough rather than striving to be of the highest quality.
As other professions, such as medicine, engineering, and architecture, promote and insist on the highest quality standards, librarianship should be no different. A defeat of metadata as a tool for information discovery will mean that “just good enough” full-text searching will replace the high-quality discovery ability that metadata can provide.
| Author Information |
| Jeffrey Beall is Catalog Librarian, Auraria Library, University of Colorado at Denver and Health Sciences Center. We welcome opinion pieces for BackTalk. Please send them to LJ/BACKTALK, 360 Park Avenue South, New York, NY 10010; fialkoff@reedbusiness.com |
