Jennifer Zaino
SemanticWeb.com Contributor

What implications does the semantic web have for enterprises?

Kingsley Idehen, founder and CEO of OpenLink Software, has some interesting thoughts on this question. His company makes Virtuoso, technology that allows enterprises or individuals to take their existing data — whether in XML, SWL, Web Services, HTML or other formats — and convert it on-the-fly to structured data. With the software in place, organizations can expose the data from all these sources as structured data using RDF (the Resource Description Framework), the general framework for describing a site’s metadata. Virtuoso is also a database that supports SPARQL, the query language of the semantic data web, for getting access to billions of interlocked items of data.

Jupitermedia recently talked with Idehen about how the semantic web changes everything.

Q: Why should enterprises care about the semantic web?

Idehen: Bill Gates touted the idea many years ago of having information at your fingertips, being a mouse click away from relevant information. But in reality you were kind of able to get to relevant information if you were ready to be locked into a report writing or business intelligence tool. Often that would lock you into an operating system or a database, which meant there was never really any chance in hell of being able to truly do this – that, say, if I receive an email, or have a conversation, that I could find all relevant information about that customer or a competitor just one click away. Then the web came and you were kind of one click away from a document that was to some degree related [to the information you were looking for].

But for corporate information being at your fingertips, that has been mercurial or mirage-like in nature at best. What the semantic data web allows is to genuinely create a so-called 360-degree view of the enterprise so all relevant data in the enterprise is one click away. You will have a URI (Uniform Resource Identifier) that identifies every customer, a URI that identifies every product in your product portfolio, a URI for every employee, a URI for every piece of doc they produce. All that means is that the systems in the enterprise — whether CRM, accounting, marketing, distributed collaborative applications, blog systems — all the data produced during these activities will have URIs so that individual pieces of data begin the drill-down process between that item of data and all related data.

A Typical scenario — a new employee gets email from an established customer who they are unaware of, maybe AT&T. It would it be nice if they simply clicked on AT&T and could learn that this was an important customer, the last transaction was this amount, and so on. That is an example of AT&T as the customer of the company having one URI, and because they are exploiting semantic web technologies all the data about AT&T is already in linked form. So that by clicking on the URI of AT&T you can see how many support cases there were, what the last kind of communications with AT&T were about, what are its latest financials.

Q: But isn’t that what you can expect from, say, a CRM system?

[What's different is] that it should be a combination of your internal data sources and external data sources. There is the conception superficially that you’d think you should do that with a CRM system. But here’s the point – you shouldn’t have to do it inside the CRM system. You should be able to do it anywhere. It is to take this issue of data access outside the confines of any application… Because if the CRM vendor doesn’t integrate all the data you need integrated, the integration can’t take place. The Idea [of the semantic web] is to empower everyone to mesh data as required by specific needs.

Just as you have the real-time enterprise, you are going to have the real-time individual. The common characteristic is this ability to discover and integrate relevant data in relevant domains.

Next Page: Building Enterprise 0.0…

Back to Page One

Can you talk about how meshing data differs from data mashups?

Idehen: Today people talk about a web of user-generated content and mashups. When you use the term mashup, it simply means using brute force to integrate the data. The defining characteristic of where we are is you mash up a Google map and overlay on it data culled from number of source. You can look at the mashup but you can’t say from that mashup I want to derive my own specific mashup. You don’t even know where the data came from for the most part and if you do you see that behind the scenes there is a lot of programming behind it.
We want meshups: I have data from here about AT&T and data there on AT&T. All that data came from a URI and was put in a semantic data web-aware environment. And because one of the characteristics of RDF is that from a single URI I can discern that AT&T is an organization — that it has a certain stock market symbol [associated with it, and address, and so on -- and the user can use that information in a mesh-up on a map]. But my data source is visible to someone else who may want to present the same data differently. You can do that with meshups because the data source is visible to you and the data is in a structured data format. It is there to be recombined, to be linked. This process is what we call linked data. You produce data in the semantic data web with linking in mind.

Q: What repercussions does the semantic web have on existing applications?

It is in terms of delivery and it is not invalidating the requirement for structure. Remember the CRM vendor says my application provides a structured view of some domain and data there. We say the data in that domain should be available through what we call a semantic data web view. So the CRM system isn’t invalidated, but it just doesn’t define my ability to work with its data anymore. But it still provides value because it provides a structured mechanism for doing data entry. You don’t want to do data entry ad hoc but there’s no harm in having ad hoc views of data.

We say that ad hoc views of data should be done …wherever you choose and you combine it with other views to create even richer views. Linked data is the defining attribute of the semantic data web.

Q: This all sounds great, but are enterprises ready to buy in?

What I’m describing marketers would like to call Enterprise 2.0 or Enterprise 3.0 but in reality it is Enterprise 0.0. What I mean is that all enterprises from the beginning of the modern IT era tried to do this but technology never delivered. …This is something they have wanted to do and given up on, and so tried all these alternatives.

What I described is precisely what has been used as the cornerstone of the value proposition for many years for relational databases, report writers, executive information systems, and more recently enterprise information integration, SOA and so on. They’re all trying to do this and none really do it the right way and none really addresses the fact that this is an issue of how you represent the data in the first place. The format the data takes is a vital component of the linkability of data.

Wide adoption takes place when people begin to see the practical dimension of the benefits of this technology. Until about 12 months ago there was a huge shortage of that. But practical demonstrations are beginning to crystallize and become available to experiment with, and the ramp-up to the semantic web will be much faster than anything else that has happened on the web.

SemTechBiz is Less Than 2 Weeks Away

The Semantic Tech & Business Conference (SemTechBiz) is coming to San Francisco on June 3-7! Join us for case studies, innovative panels, tutorials, and keynotes that will provide you with practical advice, hands-on guidance, and breakthrough approaches to solving business problems with semantic technology. Passes go up $200 at the door. Sign up now and save !