"The emerging Semantic Web will require us to dramatically rethink traditional notions of how business, data/information, application, and technology architectures are conceptualized and realized within an enterprise."

Opening Statement

Our two-decades-old World Wide Web architecture is long past due for an upgrade. During what we might call the "Web 1.0-2.0 epoch," demand for computing has grown across every enterprise, in every sector, around the globe. We continue to struggle to meet this demand using our traditional approaches to building and managing enterprise information systems. Mounting barriers of complexity and scalability continue to hinder business agility, increase costs, and constrain overall productivity. The Semantic Web (also often referred to as Web 3.0) is emerging as the prescribed solution, and it offers us a compelling architectural framework upon which to build next-generation Internet-ready applications.

THE SEMANTIC WEB ARRIVES

Today's Internet consists of a coarsely woven fabric of hypertext links among billions of largely unstructured Web pages, all generally designed to be read by humans. The data behind these Web pages is selectively shared, but the semantic definitions either (1) don't exist, or (2) are locked away behind firewalls within our enterprise systems. This is so much the case that, across the Internet, we've evolved an entirely new architectural integration layer, with point-to-point semantic interfaces and translation exchanges based on services realized through service-oriented architecture (SOA).

Therein lies the problem -- unstructured information overload, a very low signal-to-noise ratio, data subjected to human interpretation, and an associated deficiency of the required semantic precision needed to achieve a higher level of machine-interpreted cognition.

With the Semantic Web, what we're striving for is to add a layer of cognitive power to the Internet's digital gray matter -- consisting of a finely woven fabric of semantically precise, linked data that can be processed automatically, by machine-based agents, on our behalf. This is to be achieved through adoption and use of a set of emerging standards promoted by W3C for Semantic Web technologies (SWTs), including Resource Description Framework (RDF), Web Ontology Language (OWL), and SPARQL. These foundational technology standards allow for the semantic linking of data and, therefore, semantic application interoperability among semantically aware applications (SAAs).

As I've suggested in my previous research on this topic,¹ the Semantic Web is indeed already here:

• SAAs are now growing "in the wild," generating and consuming semantically linked data. Leading search vendors are embracing annotation through semantic tagging constructs (typically using RDFa or microformats). In addition, a number of automated translation techniques are emerging for RDF-izing current structured -- and in some cases, even unstructured -- data that exists in legacy formats on today's Internet.

• Open source communities are embracing the emerging W3C standards, and major commercial vendors are now adding "semantic" features or extensions into their mainstream IT management products.

• The W3C's Linked Open Data initiative is growing very quickly.² The US government is jumping on the bandwagon as well, with an increasing volume of real-time government data now being released and translated to RDF.

Thus, we may describe the era paralleling -- but also closely trailing -- today's rise of the Semantic Web as the era of the semantic enterprise (SE). A semantic enterprise can be defined as one that exploits SWTs for applications within the enterprise. We can call the underlying architecture necessary to enable the SE a semantic enterprise architecture (SEA).

SETTING SAIL TOWARD A NEW "SEA"

Although a full exploration of the principles of SEA is well beyond the scope of this opening statement, it's becoming clear to me that a revolutionary mind-shift will be needed in the way we've traditionally approached enterprise architecture. The emerging Semantic Web will require us to dramatically rethink traditional notions of how business, data/information, application, and technology architectures are conceptualized and realized within an enterprise.

This is far more than a simple terminology shift (e.g., to ontological engineering from what we used to refer to as information or data architecture). SEA will require architects to think about their enterprise from an "outside-in" perspective, where (1) open-world assumptions can apply, (2) enterprise data is largely self-describing through a tight coupling with federated ontologies (many of which are outside the direct control of the enterprise), and (3) our applications are able to effectively infer, deduce, and calculate across both private and public linked data in ways not specifically envisioned at the time they were originally designed.

SEMANTICALLY AWARE APPLICATIONS

SAAs are Internet-ready applications that generate or consume Semantic Web data -- regardless of whether the applications and semantically linked data live inside or outside the firewall. In either venue, SAAs are designed, developed, and implemented in much the same manner and will use data/information, governed by ontologies, that cross-cut both the enterprise (private) and Internet (public) domains. I suggest (and our authors concur) that in the very near future, we will find beneficial applications of SAAs within specific business problem domains where traditional enterprise systems have fallen short, including:

• Business intelligence (BI)

• Data mining (across structured and unstructured data stores)

• E-discovery (across centralized and distributed data stores)

• Customer relationship management (CRM) systems, leveraging the entirety of the social networking fabric

• Dynamic business rules (inference engine) optimization

• Ontology-based security/trust credentialing, private social networks, and public referral/viral product marketing using FOAF (Friend of a Friend), POWDER (Protocol for Web Description Resource), and other maturing standards

• Extract, transform, load (ETL) services (RDF-izers), which automatically transform and publish enterprise data into private and/or public ontological stores

• Embedded control, telemetry, and data acquisition systems relating to devices, equipment, and sensors, including (but not limited to) enablement of smart grid³ energy management systems

CHALLENGES FOR THE SEMANTIC ENTERPRISE

Of course, no amount of hand-waving, or cheerleading, will induce SWTs to solve all our enterprise computing challenges overnight. While the foundation for real solutions to today's most pressing information management problems is now here, there are certainly gaps and weaknesses in Semantic Web-based standards and technologies that will need to be addressed. These include:

• Semantic data governance, provenance, trust, and ownership across shared semantic data ontologies⁴

• Security, confidentiality, privacy, and appropriate use of data in a highly federated, distributed semantic data architecture

• Opportunities (as well as threats) related to potential monetization and commercialization of parts of the Semantic Web

• Rapidly evolving standards; vendor products to support the SE still in the inception phase of the product development cycle

I see three top action items for CIOs interested in pursuing the semantic enterprise vision:

1. Build general awareness and understanding across the enterprise for the practical application of an SEA roadmap. Invest in training and skills development in order to leverage SWTs in your specific environment.

2. Develop potential use cases for SAAs, based on your own organization's strategic business and technology opportunity areas. Charter at least one SWT-based pilot project (or proof of concept) in the coming year.

3. Begin thinking about your data/information architecture in terms of ontological engineering in order to semantically link to existing or emerging domain or industry ontologies.

IN THIS MONTH'S ISSUE

In this issue, each of our expert authors contributes to our understanding of this broad subject area through insights and unique domain knowledge. We begin with an article by independent researcher and consultant Paola Di Maio, who level-sets us in the semantics of the term "semantic" across the several different contexts in which we're seeing it applied today. She adds significantly to our understanding of what potential influences the Semantic Web will have within the enterprise.

Next, John Kuriakose of the Center for Knowledge-Driven Information Systems (CKDIS) labs at Infosys discusses some of the many enterprise applications enabled by SWTs. Kuriakose introduces us to the semantic technology stack, presents a roadmap for SWT adoption, and suggests ways to bridge the divide between traditional object/relational data stores, RDF-based triple stores, and OWL-based ontological definitions.

Our third article is by data integration expert Shamod Lacoul, who drills down into the technical case for adopting SWTs for data integration. Lacoul argues that SWTs show promise in addressing the top data integration barrier we've been struggling with for decades -- specifically, the lack of an overarching shared semantic model. He also gives us insight into graph theory, which serves as the foundation for how Semantic Web data is linked, and points out the advantages a distributed semantic data model offers in its ability to adapt to the ever-changing requirements of a business.

Next up, Cutter Senior Consultants Bhuvan Unhelkar and San Murugesan explore the overall business case for adoption of SWTs within the enterprise. Included in their article is a wonderfully succinct statement (definitely tweet-able, at just around 140 characters!) that effectively sums up the central thesis of this month's Cutter IT Journal.

By exploiting the technologies of the Semantic Web, an SE can create a people-machine continuum that enhances business agility.

Finally, MIT researchers Ken Lee and Ed Schuster introduce their Lee-Schuster Semantic Enterprise Architecture (LSSEA) in a fascinating case study in semantic engineering for the ERP problem space. While LSSEA is not strictly an SWT-based solution (absent W3C's OWL, RDF, and SPARQL standards), they argue that it is an effective way to deliver mathematical models to users quickly and cheaply, enabling experimentation and rapid adaptation.

At this moment in time, the new, transformational, Web 3.0 "cold front" is on a collision course with the muggy, evolutionary "stationary front" of challenged enterprise information and application architectures of the past 20 years. We face a potential storm of opportunity, and in this month's issue of Cutter IT Journal, our expert authors have faced this topic head-on. So let's batten down the hatches and together set sail for that new and not-so-distant shore -- the truly semantic enterprise.

ENDNOTES

¹ Ummel, Mitchell. "The Semantic Web 3.0 Mashup Universe: Coming to a Browser Near You ." Cutter Consortium Business Intelligence Executive Update, Vol. 9, No. 4, 2009.

²W3C Linked Open Data Project (http://esw.w3.org/topic/SweoIG/TaskForces/CommunityProjects/LinkingOpenData).

³ "Smart grid." Wikipedia (http://en.wikipedia.org/wiki/Smart_grid).

⁴ See "Protocol for Web Description Resources (POWDER): Primer." W3C, September 2009 (www.w3.org/TR/2009/NOTE-powder-primer-20090901).

ABOUT THE AUTHOR