Technical Background

From a technical perspective, the information model's ability to accurately express the common understanding of services is achieved with the following process:

1.  The semantic information model is built, often on the basis of an industry-standard language, such as ACORD for insurance, HL7 for health care, or FIX for financial services.

2.  The relevant services are identified.

3.  Each service interface or schema is mapped to the relevant concepts in the semantic model. This mapping starts off coarsely grained, giving proper semantic meaning to the service message as a whole (or to major message components) and proceeds to a fine granularity, assigning meaning to every field in its proper context. The flexibility inherent in these mappings allows the model's stakeholders to agree on the precise meaning of the services involved, thus eliminating the confusion caused by heterogeneous interfaces.

This is an incremental process. Identifying services and mapping them to the semantic model generally facilitates the understanding of how the business works, thereby enhancing the semantic model.

Each individual Web service typically has its own language, expressed in a schema, for input and output. When these languages are integrated through a semantic model, users get a coherent understanding of the schemas, which provides maximal transparency, integration, and IT infrastructural agility. The semantic model allows flexibility in updates to a Web service interface flexibly - clients can "understand" the changes with ease, given that the differences between schemas have been resolved by reference to an information model. When a client requires a service, the service that satisfies the business need can be discovered by reference to the semantic information model, with no need to predetermine the exact service interface. When a client requires input that is incompatible with a given service's output, then translation scripts transforming between the two are an automatic product of the semantic model and mappings, having been inferred from the semantic relationship established previously through the semantic model.

Aligning IT and the Business

IT staff understand the technical value of greater transparency, agility, and integration, but often fail to connect these to the ultimate needs of the enterprise, namely creating value according to the goals of the organization. Semantics as added to SOAs, however, tie the technical achievements tightly to the business context.

Transparency

Transparency is critical to modern organizations. Sarbanes-Oxley legislation, Basel II restrictions, and other accounting and privacy regulations in various countries make it essential to expose the true meaning of information storage and transfer to the watchful eyes of internal and external auditors.

Transparency in the enterprise is defined as the ability to trace and to understand the meaning behind enterprise data and enterprise assets, an ability increasingly important to business intelligence and executive reporting. Without a consensual understanding of a given business term, ambiguity and confusion reign. Semantics ensure that Web services and resulting user-facing reports use an unambiguous business language, and that all messages and reports can be traced back to their data sources for interpretation, verification, and auditing. Accurate business information and analysis, where every piece of data is traced back to its source and is related to its true meaning, creates competitive advantage by transparency to management. Customer-facing employees gain tremendous power to enhance their service quality, with a transparent view of a customer's personal preferences and dealings with the business.

Integration

Integration allows processes throughout the SOA to act as a single, unified process, a key driver of efficiency when communicating within the enterprise or with partners. While there are many prerequisites to achieving genuine integration, a semantically correct knowledge of IT systems (metadata) and the existence of a common business language (canonical message formats) are two of the most critical elements. Up-to-date information of IT systems and a semantic information model realized in canonical message service interfaces are two of the most critical elements in achieving sufficient integration.

For instance, a closed purchase order might require that the primary system tracking orders (e.g., an ERP system) be updated. Yet the customer relationship management (CRM), financial reporting, systems responsible for delivering the product, HR systems that track commissions, and data warehouses that track sales trends all require synchronization. Moreover, integration supports new dynamic business practices such as taking into account live sales data to automatically tune dynamic pricing or to redeploy marketing resources to maximize revenue.

Agility

Agility is achieved when semantically accurate business information ensures that service requests and responses correctly incorporate required data. When processes in SOA are modeled in terms of a common business language (instead of being tied to specific data sources), it is much easier to change business processes.

One major driver for the deployment of SOAs in the enterprise is the vision of the real-time enterprise (sometimes referred to as the "zero-latency enterprise"). The architectures are intended to allow many kinds of agility including the ability to responsively change parameters, the ability to restructure and customize business processes, the ability to prepare for organizational changes, and the ability to update, change, and integrate IT systems in response to changing business needs or new technologies.

When all business information is available enterprise-wide through a semantic information model, any field or data structure can be channeled into a service. When processes are modeled in terms of a common information model rather than tied to specific information sources or services, it is much easier to add and change processes.

Conclusion

By applying semantics, Web services break away from dependency on specific schemas and data sources. The business meaning of each service is exposed for semantic discovery. Moreover, clients can search and make calls on individual or composed services with the required functionality. Therefore, even if no one service exposes the needed interface, the client can still find the needed services, send them the required input, and transform their outputs to meet its needs.

When enterprise architects anchor their SOAs in a semantic model, integration, messaging, and other projects no longer suffer from the errors typically caused by inconsistent terminologies or definitions. As semantics boost the business value of SOA-sustaining Web services, we can look forward to a rapid growth in adoption of both semantic and service-oriented technologies.