Leveraging SOA & Web Services

SOA (Service-Oriented Architectures) and Web Services / WOA (Web-Oriented Architectures) both complement and challenge IMS - and each other.

SOA is the descendant of distributed computing and modular programming - simply a system-level, distributed computing design style, implementable in any decentralized technology such as CORBA, DCOM, SOAP, RMI, and so forth. SOA, via a "loose coupling" of technologies underlying applications (e.g. operating systems, programming languages), breaks and "de-siloes" application functionality into individual units or services which can be reused and can pass data among each other over a network as they participate in business processes.

WOA is considered easier and less expensive to implement, but less scalable (for the moment) and more concerned with the user interface at the endpoints.

Personalizing Self-Service

Increasing mobility and demands for instant gratification and service on the part of end users is spurring businesses to reach and satisfy customers via greater convenience, speed, high availability and above all, personalization. Traditional IVR systems are being replaced with newer standardsbased web-centric platforms that can provide new forms of enhanced personalized and context-based self-service delivered through voice, video, text and mobile web-based channels. VoiceObjects is a company that has liberated companies from one-sizefits- all IVRs and has pioneered contextual communications for the mobile users. With VoiceObjects technology, mobile operators possessing multiple delivery channels and multimodal applications and services can deliver information, entertainment and conversations in the proper context.

VoiceObjects has alliances to build advanced self-service solutions with such major players as SAP (now shipping VoiceObjects as part of SAP NetWeaver Voice) and Nortel (customers can purchase VoiceObjects from Nortel worldwide). Moreover, VoiceObjects is now compatible with IVRs from Aspect, Avaya, Nortel, and Oracle.

VoiceObjects' CTO and Co-founder, Michael Codini, says, "Our technology is one piece of the SOA puzzle in the communications world. We see in the financial services industry where a lot of telecom business there has been married to outsourcing services and any type of host connection being used in the telephony world is usually legacy in nature. So we are still far away from being truly integrated into a SOA architecture. That's one of the challenges we see in the overall market. At the same time companies have long time relationships on the contractual side with hosting providers."

"Specifically we see a transition going on in the IVR world," says Codini, "as it moves to an open systems architecture. But many people who are running telecom systems in a large enterprise never seem aligned with a core IT strategy at all. Sometimes you move into a building and you hope to present a great story around Web Services and SOA, but the people in charge aren't really familiar with Oracle/BEA WebLogic or anything equivalent, because they have never touched those products, so there's still quite a stretch to accomplish this in the market today."

Here Comes the Semantic Web

Something you'll be hearing a lot about in the near future will be the Semantic Web, which provides a common framework enabling data to be shared and reused across application, enterprise, and community boundaries. The Semantic Web efficiently represents data on the web, giving it the persona of a globally-linked database, making the content of web documents accessible and "understood" by computers and software. To pull off this feat, semantic technologies are used that represent meaning using ontologies and provide reasoning through the relationships, rules, logic, and conditions represented in those ontologies.

To represent the Semantic Web, you need a global naming scheme (URIs), a standard syntax for describing data such as RDF (Resource Description Framework), a standard means of describing the properties of that data (e.g. RDF-Schema), and a standard means of describing relationships between data items, such as ontologies defined with OWL, the Web Ontology Language. (OWL is designed so that applications can process the content of information instead of merely presenting it to humans. OWL enables greater machine interpretability of web content than that supported by XML, RDF, and RDF-Schema by providing additional vocabulary along with a formal semantics.)

A pioneer in this area is Thetus Corporation. Its flagship product, Thetus Publisher, provides a comprehensive framework for semantic knowledge modeling, searching, sharing and tracking knowledge across the enterprise. Thetus semantic solutions empower various kinds of data-rich organizations to leverage their knowledge to drive complex analysis and informed decision-making.

Thetus' Vice President of Marketing and Services, Philip Pridmore-Brown, says, "We're a six-year-old company that's growing its customer base pretty rapidly, including customer deployments around Semantic SOA. Our first product line was Thetus Publisher, which has been shipping for four years. In introducing the Semantic Web concept, many core technologies have been developed out of that, such as OWL, various rule standards and inference engines. Our view is that we take advantage of those components of the technology base in a very pragmatic way to try and go about solving our customer's problems. Thetus Publisher, our core product, incorporates many of those elements and takes advantage of those standards."

"In a sense, we're more of an infrastructure company than anything else," says Pridmore-Brown. "Although in the last two years or so we've been moving more toward the solution space in recognition that customers really want things that cover more of the spectrum from the core infrastructure all the way to the user experience. Typically, our customers are people who deal with fairly diverse datasets. Historically, much of our activity centered on analysis products working with such things as 'fusion centers' where one melded together data and services, and doing higher-level analysis on that, then generating information products or analysis products."

"Since early 2007 we've seen a pretty big upswing in interest in what's called 'Semantic SOA'," says Pridmore-Brown. "This trend really reflects where many people are with their enterprise infrastructures. One thing we've noticed, particularly with our federal government customers, is that they've spent much of the past two years putting together data services and actually understanding how they can expose different data sources and make them accessible.

The logical follow-on to that is that now that I can access this data, and I can interact with it, and ask, 'What does it mean and how does it fit into an overall conceptual model?'." That's a gap that we can fill."

"We deal quite a bit with federal government agencies that work on intelligence and defense challenges," says Pridmore-Brown. "They're particularly interested in being able to adapt to changing situations. This ability to dynamically stand up combinations of data and services and configure those, and be able to know how to put those together, is a very important aspect of their business process. Typically when you do that, many of these data services come from external sources or applications where you have no control over the schema or the format of what you're receiving. So, having a model that allows you to say what that means and where it fits is important and it's what they're using the data technology for and, more broadly, semantics in a SOA environment."

"A big part of what we do is to provide the ability to in a sense match the user context with the application and data context," says Pridmore-Brown. "So whether they're data services or communications services, it's important to be able to say what those services are capable of, not so much from an API level, but where they fit in the overall context of what the user is trying to accomplish and the available data sources. So basically we can take those and assess the context of the user, and determine how these things fit together, and then also being able to filter out 'noise' or unnecessary data which is of no interest to people. So, the model that we manage and that you can define in Thetus and interact with, enables you to define both the user context and your data or information architecture context as well. We stress the notion of being able to assess a fit between a disparate set of applications and services, and then assess the fit of the user context against those. The answer that you get depends on the context in which you're asking. That's a core component of a Semantic SOA architecture."

"In most cases the goal is to assemble what people loosely term 'composite applications'," says Pridmore-Brown. "These apps can be a combination of different data services, application services, and everything ranging from communications channels to visualizations and the underlying data assets that feed those things. Actually, when you consider this ability to use the semantic model to define how these things fit together, when you look at any of the what I'll term 'big players' in the enterprise application space such as SAP and Oracle/BEA, you can see that they're all looking at semantic models and technologies to help bridge this gap. SAP is one of the leaders. They're investing a tremendous amount into research and development of this type of technology. It's another trend we're watching. Semantic technology is being exposed in things such as IBM Sharepoint and other enterprise platforms that reflect much of what has already been developed at the core IT infrastructure and it's been used from behind the scenes, but it's starting to make its way up to the user level."

"Our Thetus platform is designed to be part of an application server infrastructure," say Pridmore-Brown, "part of a SOA. It allows you to define and manage those models. One core thing we do is to track the history of them, so that you understand how they're changing over time. By understanding your 'change trajectory' over time, it also enables you to predict where you need to adjust items in the future. So you can do some level of projection based on those models. We also can do automated classification and searching based on the inference rules built into the standards that come with OWL that we support. With sparse or inconsistent information about a service or about a piece of data, you can place it based on these inferences and you can bridge some of those gaps."

"The other core piece of managing these models consists of policies," says Pridmore- Brown, "because, depending on who you are and what rights you have, you may see very different views of the model and policy offers help drive context. And context is of course linked to policy in that, depending on your role and your permissions, what we can give you in terms of capabilities will change. For example, depending on where you are and who you are, if you ask for the best possible data about a person or the best possible answer to a question, we can manage all of the policy rules concerning that, particularly in environments involving classified information."

Richard Grigonis is Executive Editor of TMC's IP Communications Group.