Semantic Integration Is Emerging Field - Should You Care?

Loraine Lawson

No doubt, you've heard of the semantic Web and semantic technology in general. But have you heard about semantic integration?


Stephen Lahanas, a principal consultant and co-founder of Semantech Inc., offered this explanation of semantic integration at Semantic Report:

"Semantic Integration represents a specialized field of practice dedicated to using Semantic Design Principles, Methodologies and technology as a facilitating mechanism (often alongside SOA) to help solve enterprise-level problems for IT."

Obviously, this is an emerging field. Lahanas admits this, cautioning that while it uses the same principles as the semantic Web, it is not the same thing.

(If you have not heard of the semantic Web, or you find it a bit confusing, start with this piece from Science Daily. It's brief, yet it offers a great explanation of semantic technology and how it would improve on our current search methods.)

Semantic integration is still a pretty obscure topic, at least in integration circles. But I'm starting to hear a bit more about it lately. Lahanas now even writes a blog devoted to the topichttp://www.semantic-conference.com/publications/#SemanticWave.

But much of the information is focused on semantic integration as a niche. I've had a hard time locating anything that really puts semantic technology or semantic integration into the broader business technology perspective.

Finally, this month, Baseline Magazine took it on as part of its broader "Understanding Semantic Web Technologies" coverage.

Baseline is one of my favorite technology publications, and this article is a great example of why. First, it does a good job of explaining why we need semantic technology, with a mindful eye on the maturity of the products. Unfortunately, it only briefly mentions semantic integration, but the piece is still helpful to IT and knowledge workers trying to get a handle on this emerging field.

The article confirms what I had expected -- even though semantic technology has been around for a long, long time, it's still far from mature. But it's not as immature and bleeding edge as you might think; in fact, Gartner believes it will emerge as one of the 10 most disruptive technologies in the next four years.

If Gartner's right, then within the next four years, semantics could move from something most people barely understand to changing the way we search information, store data and, of course, approach data integration.

It's a feature article, and therefore covers a lot of ground. However, here are the main points I think are most useful if you're trying to assess how and if it matters to you:

  • If you're a drug research company or work in the life sciences, then look into semantics. As the article explains, these companies could most benefit from semantics, because data hidden in obscure surveys -- often in archives - can lead to major breakthroughs in your work.
  • Law firms, the banking industry and other companies dealing with massive amounts of court or government documents or intelligence information are candidates for early adoption. Fraud detection firms are also embracing semantic technology, because it can help find connections that ordinary search would miss.

Even if you think you don't need semantics now, start familiarizing yourself with the terms and structure that comprise semantics. At some point, this is something you'll want because it's so much closer to the way we want to find information.


First thing to know: Semantics is all about ontology. According to the Semantic Integration blog, ontology is the level above taxonomy. It includes vocabulary and taxonomy and "represents a structure that expresses both a hierarchy and a set of relationships between vocabulary 'elements' within that hierarchy." If you think creating a taxonomy is pain in the neck, brace yourself.


Second, if you wait, semantics will probably become simpler. Some people believe it must.


But, and this is the third thing you should know, one possible casualty on the road to simplicity may be your beloved relational databases. Some say relational databases just aren't appropriate for storing data and operating in the way that supports semantic integration and search.


I intend to follow up on this topic with a Q&A in the near future. Feel free to e-mail me your questions.

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.


More from Our Network
Add Comment      Leave a comment on this blog post
Aug 12, 2008 11:58 AM Josh Josh  says:
Really interesting stuff. Thanks! Reply
Aug 14, 2008 3:08 PM Samir Kumar Mishra Samir Kumar Mishra  says:
Yes we need to care about Semantic integration otherwise we will end up with lots of junk data that carries no importance whatsoever. Today the major problem we are facing is data integration and that is the result of "Should We care?" attitude developers been carrying around the world for last 6 decades.If we don't care now then we will not be able to get anything out of this ever growing pile of data.My research is based on the same topic and you may find few points I discussed on my blog as well.CheersSamirhtt:p://semanticdb.blogspot.com/ Reply
Sep 25, 2009 8:29 AM Peter Rajsky Peter Rajsky  says:

Nice to see article about semantic integration. Semantic integration is broad term. It could solve many problems and I really believe it will be useful in our everyday life of EAI architects/designers/programmers. Unfortunatelly research in this area is not oriented to practical problems in EAI. But I think there are two areas, where it will play important role in the "near" future (but I do not believe that in 4 years):

1) Explicit service/API contract - The goal is to make API as explicit as possible (see more on http://it.toolbox.com/blogs/system-integration-theory/learn-make-explicit-analyze-and-monitor-impact-fix-34224). Semantics is crucial for contract definition.

2) RDF based messages - RDF is better than pure XML, because it is about concepts and not about data structures. It is easier to integrate data from many sources (everything is triple - <subject, predicate, object>). But the are still important problems, why we not use it in practice (see http://it.toolbox.com/blogs/system-integration-theory/canonical-data-model-using-industry-standard-data-models-26635 below).

I think (1) is more important. The problem is that there are no tools at the moment. There are various reasons. E.g. People are lazy to create explicit syntactic contracts, so they do not need tools for explicit semantic contracts. They do not need event contracts (see whole REST in EAI movement)...


Post a comment





(Maximum characters: 1200). You have 1200 characters left.




Subscribe Daily Edge Newsletters

Sign up now and get the best business technology insights direct to your inbox.

Subscribe Daily Edge Newsletters

Sign up now and get the best business technology insights direct to your inbox.