While designing a new CMS implementation we wanted to really get it right. That meant outputting strict xhtml, css formatting, trying to adhere to accessibility guidelines, etcetera. Of course, the issue of metadata came up. What metadata would we render to the web pages? And in what format? Which is what got me looking at Dublin Core.
For the un-initiated, Dublin Core (DCMI, dublincore.org) is a set of elements and attributes that can be used to define the data (it is, therefore, metadata). It can be added in different formats – on the web usually as META elements. Dublin Core metatags are easy to spot if you browse through the source of an html document – they all start with “dc.”, e.g., “dc.author”.
Now why would one want to use DC tags? According to the DCMI, “Dublin Core metadata is used to supplement existing methods for searching and indexing Web-based metadata”. Which doesn’t really say much but sounds vaguely promising. So how exactly would we benefit from using DC? Well, even the DCMI’s website only lists who could benefit, without exactly pinpointing how. There isn’t a whole lot of information about the ways DC could actually help us search and index web-based metadata.
Would DC help inform search engines about the contents of a webpage on our sites? Well, no. Actually, I haven’t found a public search engine that even reads DC; “spamming the index” has lead most of them to ignore html metadata and Google, for example, has never even supported html metadata, apparently only reading “title” and “description” of the non-DC variety. Of course, you could use the DC metadata in your own site search engine. But then again, why would you? It would probably take a lot more configuration than just using the metadata from your repository, be it SQL or XML.
So maybe it’s great for interchanging repositories? If you have lots of different collections (web documents, book abstracts, multimedia files etcetera) in different systems, or different organisations, and have to somehow intertwine those, it would be great to have a “standard” set of metadata to read. Unfortunately, DC is as restrictive (it’s basically a set of only 15 different elements) as it is not: for example, the “dc.date” element contains “a date of an event in the lifecycle of the resource”. This leaves so much room for interpretation and specification in different organisations or systems you’ll end up having to set standards within DC for them. Which somehow, to me, defies the purpose of using DC (why not set your own standard straight away then, without the restrictions, and without the added muddle of a “dc.” prefix?).
Of course, I’m not trying to defeat the use of Dublin Core metadata with a short blog post. My question is sincere: what’s the use of Dublin Core? If anyone has answers, I’d love to hear them!
This post was originally published on the blog of Content Management Professionals Benelux. I have added the comments received on the original post as one, below.
COMMENT BY BOB DOYLE, 4 SEPTEMBER 06
Hi Adriaan,
The short answer is – if you have need for any of the metadata defined by Dublin Core, you should use their pre-existing standard types and resist the temptation to define your own.
Don’t fall into the “not invented here” trap just because you are smart and can do so. Standards promise greater interoperability in the future than most of us can see at the present time.
For example, the Information Architecture Institute created a Resource Library.
http://www.iainstitute.org/library/
It includes tags like author, date, title, subjetc, etc.
Now IAs are mostly all trained as librarians, with masters degrees in library and information science. They, of all people, should know the benefits of the Dublin Core set of tags.
But they made up their own attributes in the Movable Type CMS-Lite they are using.
At CM Pros, we tooks away the good parts of their idea for a Resource Library, but our interface encodes the same tags as dc:author, etc. Indeed, we went to a full RDF specification, which sadly makes our library a bit slow.
Now we make use of our tags – we can display a list of works by author, etc.
If you don’t use metadata tags in some way, allowing the user to navigate via the tags, they are waste of your design time, of content contributor time, and of metadata editors (Bob Boiko’s “meditor”) time.
If you do use your metadata tags, you can help your users find the 10-20% of information in your content not easily findable by search alone.
I wrote about this in an article – “Metadata, Think Outside the Docs”.
http://www.econtentmag.com/Articles/ArticleReader.aspx?ArticleID=7947
Hope this helps,
Bob.
COMMENT BY ADRIAAN M. BLOEM, 4 SEPTEMBER 06
Your comments about metadata in general are excellent – but with regards to using Dublin Core, I’m not convinced (of course, as usual, exaggerating the point for the sake of debate).
To me it appears DC falls in the gap in between plain html – which took so much designers’ abuse it never fulfilled its semantic premise – and Tim Berner Lee’s “semantic web” – which is largely still a future promise. As such, it was never widely adopted and I fear its time has passed without truly being in the spotlight. That means future metadata interoperability probably will be solved in other ways – XSL translations, maybe aided by RDF dictionaries, etcetera. I suppose those translations wouldn’t even be easier if all of them were to be DC compliant, as there’s so much room for navigation within that metadata set.
Which would make your point about metadata in general hold true even in that respect: if you’re not using the specifics of DC per se, why would it justify the effort of striving to comply with that standard?
COMMENT BY DMITRY, 5 SEPTEMBER 06
DC and other types of metadata are great for switching to “subject centric” organization of information. For example, if we assign “DC.Creator” for all our documents, we can find all documents created by a specific person. If we add department responsible for document publishing as “DC.Publisher”, we can collect all documents published by specific department. If we add “main subject”, “subject area”, “document type” etc. metatags and combine full-text search and faceted navigation based on these “axes”, we have a very good chance to increase findability. Using metadata helps also to connect documents and subjects and implement portal-like pages for main subjects (people, companies, products, tasks, services, projects, business processes etc.)