Corporate use of Dublin Core tags

June, 2002

Background
Tags (also called "metatags") are labels inserted into electronic documents to facilitate manipulation by software programs. They are commonly used to format documents (e.g. apply font styles and sizes to headings or paragraphs) and to improve the effectiveness of search engines. A "tag" consists of two parts -- a label (e.g. <Title>) and a value (e.g. <Corporate use of Dublin Core tags>).

In order for web browsers and commercial search engines to use tagged information, the tags have to be standardized. The Dublin Core set of tags has become the de facto industry standard for describing bibliographic information. The issue here is which search engines can recognize and use Dublin Core tags and how should knowledge base publishers use them for best results.

Original question
This question was initiated by a Society member interested in how others are using the Dublin Core tags. Specifically, she wanted to know:

1) What search engines read Dublin Core? Are there browse tools you recommend? We've seen the tools list on the site but understand that commercial tools such as Inktomi read Dublin Core and are looking for best practices. Also, do you use HTML tags in addition to Dublin Core tags on your public web site so that all kinds of search engines can access the values in the tags?

2) How have you seen other organizations use Dublin Core to tag information with content management systems and outside content management systems and then leverage by integrating browse and search tools that read Dublin Core.

3) How is the Dublin Core tag dc:subject used within organizations? Is it used by selecting from a controlled vocabulary that is a subject taxonomy? Are the values from other tags rolled up into dc:subject at the publishing point so search engines read just the dc:subject tag (converted to the HTML metadata keywords tag)? What is the recommended use?

4) How are organizations managing metadata/taxonomies in terms of synonyms, related terms, etc.? Is there a recommended thesaurus management tool to work with Dublin Core? Is there a Dublin Core tag that can capture related terms, synonyms, etc.? We are thinking about the management of the relationships between terms.

Member responses
1. According to the UKOLN project at the University of Bath, subject headings will ideally come from a controlled vocabulary of the enterprise or institution. A small utility program, DC Assist, describes each of the Dublin Core tags, describes "best practices" if any, and gives an example of its use.

2. We use a subset of the Dublin Core along with some "in-house" tags to help with records retention and copyright. For the subject tag we use a Taxonomy. Inktomi does support metatags and has several powerful features that can be used to enhance Intranet search for sites that use metatags. We are interested in finding out what other companies are doing with metatags -- especially the subject tag and the management of taxonomy terms. If there is anyone else working these topics we would like to talk to them.

3. The State of Minnesota uses Dublin Core metadata with the Inktomi Search Content Classification Engine. You can see the results of the BRIDGES project, an effort to provide access to environmental information from multiple state agencies. This page contains a link to their metadata training guide (PDF). Eileen Quam, the person who led the BRIDGES metadata project will be a guest expert at the Montague Institute "Taxonomy integration" roundtable in October.

Montague Institute's experience
We use the Coveo search engine with a subset of the Dublin Core tags for our site search. The tags we use are:

Title
Creator (author)
Subject
Description
Publisher
Date
Identifier
Rights

The metatag values we assign are drawn from a relational database that includes document information (i.e. author, title, identifier), a controlled vocabulary, and a thesaurus. We use a small subset of the vocabulary terms as metatag values, and we ignore thesaurus terms. Vocabulary terms are entered in the dc.subject metatag, which Inktomi uses to select documents for its category search. We do not use metatags to maximize our exposure on public search engines, as some people do. Our primary goal is to make it easier for people to navigate our public and member web sites.

Our complete vocabulary and thesaurus is used on the A - Z Index page, which draws data from our Knowledge Base on the web. We view the Inktomi site search and the A - Z Index as complimentary.

We will soon be testing the Inktomi thesaurus capability, which allows users to expand their search based on synonyms or related terms. For the test, we'll export the thesaurus in our Knowledge Base as an XML file and make it available to Inktomi.

Other search engines
Other Dublin Core-capable search engines suggested by members include:

SWISH-E
Verity
HiSoftware (see the list of white papers and case studies on this page)
Convera

See also
This 1999 article by Kelly Doran describes in detail how Weyerhaueser uses a modified Dublin Core set of metatags with the Verity search engine.

Kelly Doran, "Metadata for corporate intranet," Online (vol. 23, no. 1), p. 42.

Created on June 27, 2002 | Updated on September 13, 2012