David S. Read

I attended my first semantic web conference this week, the Semantic Web Summit (East) held in Boston.  The focus of the event was how businesses can leverage semantic technologies.  I was interested in what people were actually doing with the technology.  The one and a half days of presentations were informative and diverse.

Our host was Mills Davis, a name that I have encountered frequently during my exploration of the semantic web.  He did a great job of keeping the sessions running on time as well as engaging the audience.  The presentations were generally crisp and clear.  In some cases the speaker presented a product that utilizes semantic concepts, describing its role in the value chain.  In other cases we heard about challenges solved with semantic technologies.

My major takeaways were: 1) semantic technologies work and are being applied to a broad spectrum of problems and 2) the potential business applications of these technologies are vast and ripe for creative minds to explore.  This all bodes well for people delving into semantic technologies since there is an infrastructure of tools and techniques available upon which to build while permitting broad opportunities to benefit from leveraging them.

As a CTO with 20+ years focused on business environments, including application development, enterprise application integration, data warehousing, and business intelligence I identified most closely with the sessions geared around intra-business and B2B uses of semantic technology.  There were other sessions looking a B2C which were well done but not applicable to the world in which I find myself currently working.

Talks by Dennis Wisnosky and Mike Dunn were particularly focused on the business value that can be achieved through the use of semantic technologies.  Further, they helped to define basic best practices that they apply to such projects.  Dennis in particular gave specific information around his processes and architecture while talking about the enormous value that his team achieved.

Heartening to me was the fact that these best practices, processes and architectures are not significantly different than those used with other enterprise system endeavors.  So we don’t need to retool all our understanding of good project management practices and infrastructure design, we just need to internalize where semantic technology best fits into the technology stack.

Stephen Wolfram gave a great talk that delved into the thinking behind Wolfram|Alpha as well as Mathematica.  The man is clearly brilliant and presented with a laid back style that showed his mastery of the subject and deep thinking within the problem domain.

I found it fascinating that Wolfram|Alpha benefitted from Mathematica and now, with the latest Mathematica release, that software package will benefit from Wolfram|Alpha.  This symbiosis, Mathematica supplying functions to solve an enormous number of problems and Wolfram|Alpha’s ability to convert a natural language query into suitable input for Mathematica, creates an ever more powerful platform for people to solve hard problems.

Although Stephen indicated that semantic concepts play an important role in the implementation of Wolfram|Alpha, the constructs used are not the semantic standards such as RDF and OWL.  He shared that these were not ready in time for his needs so he built his own representations and algorithms.  Still, that the concepts are similar shows a logical value to the technology separate from a given representation.

At this point I think it is safe to say that businesses considering any type of application development, enterprise integration or BI initiative, including construction of a data warehouse, should stop and evaluate how to effectively leverage semantic technologies. Why?  Principally, because the underlying data structure for semantic information doesn’t require a fixed set of relationships beyond the triple. Change at the data level is painful and that issue can clearly be addressed with semantic technology.  There are other promises, but this one has already delivered.

The phrase that was used several times, and clearly demonstrated by several use cases that were presented, is that with traditional systems “change is expensive” while with semantic technologies “change is cheap.” Although I see the basis for the claim, I’m sure it can be overstated.  Like any technology this one has place(es) where it will work well and places where it will make the task harder than other options.

We certainly do not know all the best practices for semantic technology use and by extension we don’t know everywhere that it might not work well.  However, semantic technology’s basis is meaning.  Since interacting with data requires discerning its meaning (often hardcoded into applications) it makes sense that using semantic technology at the backend will empower more effective use of the data.

There is much to explore, particularly around leveraging the flexibility of semantic data relationships, within this technology.  My personal focus will continue to be in that realm as well as understanding where work is needed to make its use more mainstream (e.g. commoditized).

Tags: enterprise applications, enterprise systems, Information Systems, linkedin, semantic web, semantics, system integration

This entry was posted on Thursday, November 18th, 2010 at 00:29 and is filed under Architecture, Information Systems, Semantic Web, Software Composition, Tools and Applications. You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.