Why bother with semantic technologies?


Semantic technologies are a key component in the move towards a better web, but as usual emerging technologies have a much slower uptake than you might think if you listen to analysts and the industry press.

It takes time before they become part of the standard toolbox of ordinary organisations and this is also the case with semantic technologies, which have so far left practitioners stuck with limiting approaches like Dublin Core and clunky content management systems.

Dublin Core comes up short

Dublin Core originated in the library world, with a focus on describing resources. The vocabulary contains a fixed number of terms, that represent what is typically found on library cards, and provides ways to express metadata like author, subject, language, etc. This has been used as a format to provide metadata for web resources, like web pages, video files, photo and images.

But when it comes to the needs of today, we experience two major limitations of Dublin Core:

  1. We now want to annotate contents by sprinkling metadata inside web resources, embedded in flexible ways. Dublin Core was not really designed to support that kind of use. As an example, think of an encyclopedia. From a librarian point of view, what you keep metadata about are the individual volumes of, say, Encyclopedia Britannica, resulting in a set of around 20 metadata records. But what we see in Wikipedia is a lot of metadata, not only about each article, but also multiple metadata records within each article. Dublin Core was not designed for this.
  2. The fixed terms of Dublin Core prevent us from expressing metadata for most interesting domains. E.g., we want to talk about dates/times, and places, and organisations, and costs, etc. And this is where we see our present-day needs going beyond what Dublin Core can provide us with.

Powerful semantic web technologies can be challenging

Technologies for the semantic web has been with us for more than ten years, and standardised RDF and OWL for nearly 8 years.

These technologies aimed to be "super Swiss Army Knives", sufficient for all kinds of needs. And they are powerful. But practical uptake has not been rapid. There were expectations that we all would quickly take the road to full semantic web content, but those expectations have been scaled down.

There are two main reasons:

  1. Tool support has been slow to emerge, and associated methods and process have not yet stabilised
  2. There has been much confusion about how a content provider should select suitable areas for semantic modelling, as well as uncertainty about how to make a business case for large scale adoption of semantic web technologies.

We have seen many early adopters of advanced semantic web approaches, but mostly by large organisations with well established information models about their domain. They typically see great value in the reuse of content that semantic web technologies offer. So there are many examples of successful use of powerful semantic web tools and techniques. But for the majority of content providers, uptake of full semantic web technologies has been slow.

Why really bother with semantic technologies?

Think about the Web 2.0 concept. Initially, the reaction was typically "Cool thing, yes, but it is not of any use to us or to our customers". A few years later, all content providers were moving in that direction. Partly because practical tools, frameworks, and platforms became available, and partly because users' preferences clearly indicated that they wanted more than just see ordinary static web content presented on their screens.

The same thing is starting to happen with semantic technologies. Simple tools offers ways to enrich content in a gradual manner, avoiding the need to do large scale modelling of application domains. And by simple means, one can provide the users with more meaningful information and interaction.

The browsers deployed around the world are becoming much better platforms for rich content. The insight is that we are actually seeing a rapid growth in the use of sematic technologies, mainly in the use of light-weight technologies that are simple to use, and that provide concrete value to users.

But this use is mostly not talked about as application of advanced semantic approaches. Rather it is seen as a pragmatical step towards better contents on the web. It is often talked about as "embedding data in web pages", and from that point of view, it has been seen as an evolutionary step taken by an organisation, not a revolutionary step.

The growth in small-scale use of semantic technologies will of course ultimately contribute to later adoption of more powerful semantic technologies. But we do not have to commit ourselves now to heavy technology. Start small now, find the low-hanging fruit, enrich content gradually, learn by doing, ... and then you will be better prepared to tackle more ambitious goals for your web presence.

Olle Olsson is senior researcher at the Swedish Institute of Computer Science (SICS), a national research institute in Information and Communication Technologies. He is also manager of the Swedish W3C Office, a part of the global World Wide Web Consortium. You can meet Olle at the Aarhus 12 web conference in November, where he will speak on how recent approaches to metadata is improving the Web

One Response to “Why bother with semantic technologies?”

  1. Fredrik says:

    Dear Mr. Olsson!
    Even though your post is now almost a year old, I could not have agreed more today. We have experienced the same shift in directions of more light-weight technologies that are simpler to use, and that provide concrete value to users. I guess the hard core semantic web people would not agree when I refer to OData and Microdata as examples of such light-weighted semantic technologies. But If you look at what’s likely to come out with OData V4, it’s getting pretty close to the concept of global unique identifiers, just that the barriers to deploy this are much less than for example with RDF and SPARQL. We hope that innovations around such light-weighted semantic technologies (i.e. OData), will be the tipping points for great initiatives like Open Data and LOD.

Leave a Reply

Most popular posts from our blog
August 12, 2009 by Janus Boye
Selecting the right CMS is not an easy task with; there is in excess of 1,000 vendors in the very dynamic CMS…
February 16, 2010 by Janus Boye
All modern CMS vendors claim to be capable of delivering content to mobile devices. Some even offer additional modules to make the…
March 21, 2011 by Janus Boye
You may be impressed by some of the features during a CMS product presentation, but in reality many of the features that…
Recent comments
April 8, 2014 by Bertrand
I would include the Apache Sling website in this list ( ) as it's what powers AEM. Understanding the…
April 8, 2014 by Scott Liewehr
Great resource, Janus. And thanks for the mention.
March 15, 2014 by John
Their site may not even be a CMS but more of a custom built web application. Also, how do you…