Letters to the Editor

Spacer

D-Lib Magazine
March 2001

Volume 7 Number 3

ISSN 1082-9873

To the Editor


The letter below was received in response to the story, "Keeping Dublin Core Simple: Cross-Domain Discovery or Resource Description?" by Carl Lagoze in the January 2001 issue of D-Lib Magazine.

Letter to the Editor, D-Lib,

The remarks by Carl Lagoze in his article, "Keeping Dublin Core Simple: Cross-Domain Discovery or Resource Description?" (Jan. 2001 issue [http://www.dlib.org/dlib/january01/lagoze/01lagoze.html]) were very interesting and he makes some excellent points in his discussion of the future direction of the Dublin Core. I completely agree with Mr. Lagoze's concern about the need for simplicity and I do not agree with the push for further refinements in DC. Still, I think this push is probably unavoidable, and some of my library colleagues--who are used to describing items within an inch of their lives--are just as guilty in this as anyone else.

But they raise an excellent point, which I don't believe is clear from the article. For librarians, "discovery" (i.e. knowing about an item's existence) is tightly linked with "description" (i.e. knowing what that item is). This is true because librarians have learned through painful experience that it is necessary to describe an item well enough to distinguish it from other, similar items. Even if there are no similar items in existence now, there most probably will be in the future. If it turns out that we cannot describe an item to such an extent, then it follows that we aren't sure exactly what we have. How then could a user prefer one version of an item to another version? The user would be forced to examine each item to determine how they differ, and in this case, we must conclude that the metadata records we have created are unsuccessful. How would you like to have to do this for the Bible? Allowing users to "discover" Bibles without providing a detailed description of which version(s) they are looking at would be a nightmare. It is incumbent for metadata creators (be they librarians, museum curators, or others) to describe something precisely enough for users to know what an item is. If we can't even fulfill this task, then how can we possibly claim that we have allowed people to "discover" an item, when we obviously don't know what that item is in the first place? So, in this way, the description of an item is absolutely necessary for its discovery. For example, the first thing a cataloger must do when a new book arrives is to determine if it is a copy of something already in the collection, or something new. (Although this may seem to be a simple task, it can become very complex) If we can't even accomplish this, the purpose of the entire catalog comes into question. Library records are designed to help us make these fine distinctions. "Discovery" without adequate "description" is therefore highly suspect, if not doomed to failure.

I agree with all of this reasoning, *except* when it comes to electronic resources. The reasoning I described above is premised on an assumption that doesn't hold true any longer: what I shall call "the assumption of permanent documents". Detailed description by librarians has always assumed that an item will never change. As a result, a bibliographic description created for a book today will remain fully valid forever (theoretically). By the way, this seems to hold true for the most part. At Princeton, we still use the descriptions from 100 years ago, many of which were copied from even earlier versions. With electronic resources and their ever-changing states, this level of description will be valid only in those rare cases when the site hasn't changed from the time of metadata creation to the moment someone uses the metadata record. A person can make a metadata record this morning, but so far as anyone knows, the website will have changed completely by this afternoon, and the information in the metadata record turns out to be bogus. Obviously, hitting upon a site that hasn't changed is only a matter of luck. In the case of a metadata record being correct for 100 years -- it is simply laughable.

As a result, I feel that the purpose and task of "bibliographic description" as we have always known it is destined for a complete overhaul, but no one is even considering this, so far as I know. It is fully understandable that people want to avoid this issue, because it's simpler to concern ourselves with other matters: "I want to find Leonardo as a 'painter-creator'" or "I think we have to have 'affiliation' in DC" instead of dealing with the vital question, "How do we guarantee that a metadata record actually gives a correct idea of the item as it is at this moment, instead of how it looked six months ago? How can we deal with this incredibly high level of maintenance?"

As opposed to description, the "discovery" points change far more slowly. For example, a website will probably not change its subjects very quickly, whereas the URL/format/publisher/dates/etc. will change constantly. Finally, coming full-circle in my argument, "description" is just as important as "discovery," but what that "description" should be, along with its entire purpose, is still very much an unknown. For now, "discovery" is the best and most fruitful area for the task at hand.

James L. Weinheimer
Princeton University


D-Lib Magazine welcomes letters related to digital library research and electronic publishing issues. Also welcome are letters with questions or comments about the magazine in general, or about articles appearing in the magazine. Please do not send those that are primarily commercial, promotional, or advertising in nature.

Letters concerning articles selected for possible publication as Letters to the Editor will be forwarded to the article authors for response. If published, the Letter to the Editor will appear with the article authors' responses whenever possible. D-Lib Magazine reserves the right to edit or shorten letters. If you prefer, you may request that your letter not be published.

Please send your Letters to the Editor to dlib@cnri.reston.va.us.


Copyright (c) 2001 Corporation for National Research Initiatives

Top | Contents | Editorial | Opinion

Search | Author Index | Title Index | Back Issues


DOI: 10.1045/letters