A few years ago OCLC created a basic set of cataloging Meta tags called the Dublin Core (named after Dublin, Ohio, where OCLC is headquartered). This page uses some underlying Dublin Core tags, but I've implemented the tags better (using the University of Bath's DC.dot described below), at the old Gilder Lehrman Collection and the American Printing History Association (for which I volunteer) web sites.
In principle Dublin Core was (and is) a system for cataloging all sorts of information -- whether electronic web pages, images, or online books, or physical artifacts. The concept was to encourage people to catalog their own creations. In 1999, the standards seem to have settled and a minimal set of 15 elements have been chosen to describe information. So simple and so compelling are these standards that in August 1999 the Consortium for the Computer Interchange of Museum Information, CIMI, adopted a draft on using Dublin Core for describing museum web pages, electronic reproductions, artifacts, and even for catalogs! I think this is extremely cool. CIMI thinks DC is so easy that even museum curators can use it! In fact, from 2001-2002, I've seen more museum job descriptions requesting familiarity with Dunlin Core, and the Getty Research Institute has put its support behind the standard. For more information on using Dublin Core in Museums, see CIMI's document Guide to Best Practice: Dublin Core, available as a PDF file.
An important article on meta data appeared in the April 2002 D-Lib magazine, entitled "Metadata Principles and Practicalities," by Erik Duval, Wayne Hodgins, Stuart Sutton, and Stuart L. Weibel. A longish tech article, it's worth skimming for major principles and issues of meta data.
Dublin Core's home page has more information about the project. The Dublin Core Metadata Initiative's FAQ has excellent explanations that are well worth reading and pondering. There's a useful PDF help file on DC from the University of Wisconsin Digital Collections Center.
Official Description: "The Dublin Core is a metadata element set intended to facilitate discovery of electronic resources. Originally conceived for author-generated description of Web resources, it has attracted the attention of formal resource description communities such as museums, libraries, government agencies and commercial organizations."
Organizing & Linking the Structural Data: How do you organize, link and structure all the different files that might constitute a larger work? How do you relate Dublin Core data to the larger work? With HTML I embed the information. However, embedding into images, audio, video, etc. may present problems, especially when those individual files are components of a larger work. Embedding variable data into each file may also make changing that information later more difficult. One answer is METS, the Metadata Encoding Transmission Standard, developed by the Digital Library Federation and housed at the Library of Congress. More information on METS.
Automation. When Dublin Core was first being implemented, some smart librarian-programmers developed Perl Scripts (and other online programs) to allow people to generate and edit Dublin Core tags interactively. It can be interesting to see how a machine indexes your pages. It can be gratifying to make the changes online (and interactively), and cut-and-paste the finished tags into your HTML document. The best: DC.Dot a Dublin Core generator. at the University of Bath.
Note: After seeing some results of machine-generated indexing of web pages (without meta tags), which I authored for the Gilder Lehrman Collection, I decided to investigate meta tags, so that I could specify how my work was indexed by machines and presented to off-site visitors viewing those abstracts at indexes like Excite. It doesn't always work, and I had to write letters to some search engines like Excite to have them start indexing Dublin Core tags. Still, the situation is much better than 1997.
What are the ten elements from Dublin Core? This comes from the DC Reference Description 1.1 recommendation:
Okay, but how do I do it? You create or generate the tags (see the links below), and insert in your HTML HEADER, where they will be invisible to anyone but you and any machines. The catalogers are probably asking How do I handle multiple authors? Here's an example from this page:
<Meta NAME="DC.Title" CONTENT="Dublin Core Meta Tags">
<Meta NAME="DC.Creator" CONTENT="Paul W. Romaine">
<Meta NAME="DC.Creator.Address" CONTENT="romaine@pipeline.com">
Let's say there are two creators for a web page -- it has been jointly authored. You can add a second DC.creator entry identical to the first. But let's say that you want to differentiate one author as primary. Currently Dublin Core has no recognized standard for doing that. One suggested way that may work (and won't "break" any software) would be to add a period (or "dot") and the number 1 or 2 to creator, the same way that e-mail was handled above. So we might have:
<Meta NAME="DC.Creator.1" CONTENT="Paul W. Romaine">
<Meta NAME="DC.Creator.2" CONTENT="Meow M. Romaine">
This practice doesn't quite save the library notion of "main entry" (the person or entity primarily responsible for a work), but it certainly doesn't hurt to separate information. General links:
Although there are programs that generate Meta tags for PCs, why not see what a robust engine does with you tags?
[Home] [Work] [Research & Notes > Meta Tags > Dublin Core] [Personal]
Contact Paul Romaine.