Comments on Building a Metadata-Based Website

Yeu Wen — Fri, 26 Jan 2007 14:52:46 GMT

I am unable to see the graphics or figures on this page. Am I missing anything?

Girijesh — Fri, 26 Jan 2007 14:51:20 GMT

Thanks for the wonderful article, it gives readers like me a great starting point.

I do have the following questions however:
a) Of the various CMS products (interwoven, vignette etc.), which ones (if any) in your opinion could best incorporate a metadata driven approach

b) We (a leading IT services & solutions company) are currently in the process of embarking on a site architecture revamp & would like to move to a metadata driven site architecture. Could you recommend some further reading material or organizations that specialize in deploying metadata driven websites

Cheers!

ZAM — Fri, 26 Jan 2007 14:51:20 GMT

how to redisgn web using OOHDM

Vikram Chati — Fri, 26 Jan 2007 14:51:20 GMT

This is indeed a very informative article. Its one of the best articles I have read regarding website creations using metadata and taxonomies.

Just one request Brett and Anca – I would like to know if there is any book or other articles published which have more elaborate and detailed examples

jens möller — Fri, 26 Jan 2007 14:51:20 GMT

congratulations!
your article is great- in fact, i´ve been waiting for stuff of this kind for quite a while now- but should be taken with a grain of salt. explaining the need for a metadata based approach to managers, even engineers, is the hard part. maybe the following idea helps: relating products to technologies, solutions and services is an excellent basis for controlling purposes such as “technology controlling” or product lifecycle management, since a product is only the embodiment of several concepts (technology, distribution etc) that in fact are the real targets for the effects of time.
yours, jens

Graham Robson — Fri, 26 Jan 2007 14:51:20 GMT

I had one of those moments reading this article, a sort of “I’m not along in the world” one. The metadata based principles outlined resonate so well with a belief and in practice what I have been pursuing for some years now.

I build metadata web applications using WebML with a tool called WebRatio (www.webratio.com). This is web development environment that you can use to design a model-driven solution which automatically generates all the implementation code. Amazing as the tool is, the key is that I have applied a metadata based design approach to the way I build solutions.

I produce an information model (UML class diagram) comprising of business level objects like “Products, Solutions, Services, Organisations, ..” and most importantly define the relationships/associations between. In some cases I even have relationship classes between classes to better define the specific nature of inter-relationships.

This underlying information model is then exploited at the business logic level by using the metadata as a dynamic navigation schema. There is only ever one instance of anything. Page content assembly and navigation is a function of the relationships that exists between the information chucks.

If you take a look at my site www.coherencedesign.co.uk you can see a complete working implementation of this approach. All the navigation elements and page content are completely data driven. New elements and pages can be added using an associated content management application (also designed with WebRatio).

I would be happy to submit an article which explains in more detail how I am applying the concepts of Building a Metadata based Website.

Anca Mosoiu — Fri, 26 Jan 2007 14:51:19 GMT

One thing that we didn’t talk about in this article is the modeling methodology that we used. We didn’t use a “traditional” or “academic” method for the simple reason that it would have meant teaching a variety of people a number of different visual languages (UML, WebML, etc). The system that we developed was used with business owners, developers, and even users to create and describe the site.

But that’s a topic for another set of articles.

john — Fri, 26 Jan 2007 14:51:19 GMT

Most situations seem to be combinations of what you refer to as traditional CMS’s and metadata-based sites. Rarely can a metadata-based site derive all contextual information from the core content of a page, but allowing the “exceptions” to be input along with the core content can provide much of the flexibility of a traditional CMS.

Adam Gaffin — Fri, 26 Jan 2007 14:51:19 GMT

Great article except that I would dismiss CMSes so quickly. Done right, (metadata is such a political issue; it’d be easy to get a metadata project mired down forever), metadata can make a CMS so much more useful. I’m not sure I’d agree with some of your CMS “cons.” We use a CMS; we have no problem republishing our content in multiple formats and our users don’t have to know anything about file locations (they do have to be able to make some basic assumptions about the content, though, for example, whether an article is “news” or “opinion”).

Brett Lider — Fri, 26 Jan 2007 14:51:19 GMT

Sorry if it is not clear in the article, but Figure 9 (http://www.boxesandarrows.com/archives/images/042103_lide… shows the use of a Relationship Type in the Ontology used for “navigation structure”—we called this Rel Type website_hierarchy. Agreed that the structure of the taxonomy and what is displayed to users should be separated in most cases. I am not familiar with OOHDM and WebML—I will definitely check them out—but at least on a conceptual level what they are do seems like providing yet another view or facet through the core concepts.

Daniel Schwabe — Fri, 26 Jan 2007 14:51:19 GMT

I would like to point out the existence of several model-based methods for the design of web applications. Some of the most popular include OOHDM (Object Oriented Hypermedia Design Method), and WebML. While such methods have their origin in academia, they have also been successfully applied to industrial strength websites.
These methods, while not cast as using metadata directly (yet), do show that it is desirable to have an additional abstraction layer between your ontology and the actual navigation structure…
The ontology describes how the information is defined – its characteristics and relations, and the navigation structure provides a “view” over these items to support a given set of target user types and tasks.