Friday, June 15, 2007

imho Ocean Semantic.........


I've been putting together an SOA Semantic Web maturity model. I'm trying to create a reference framework that can help me sort through the confusion, complexity, and diversity of solutions/components/tools in this market.

In developing the framework, I've been working from a basic principle: SOA. In other words, SOA refers fundamentally to a paradigm that focuses on maximizing the reuse, sharing, and standards-based interoperability of key resources over distributed environments. In an SOA context, then, we can conceive of semantics (of data, services, apps, business processes, etc.) as perhaps the most important resource that must be shared. Hence the "SOA Semantic Web."

I already have a recently developed SOA framework that gets me 90 percent of the way there. It's the master data management (MDM) maturity model that is the conceptual backbone of my MDM market coverage for Current Analysis. You can see how I use that maturity model to compare/contrast MDM vendors' solution sets (e.g., IBM, Oracle, Teradata, TIBCO, SAS/DataFlux etc.) if you go to and subscribe to my Data Management module (hey....I told you I make a living somehow...this is an explicit plug for my bread-and-butter). That MDM maturity model includes an explicit notion of "governance" of this resource (i.e., master data) within a "domain" according to a "domain model." I find these notions essential to understanding how a vocabulary (i.e., ontology) is controlled within a Semantic Web environment.

To some degree, if we use the word "semantic" in place of "data" in the maturity model (and make a variety of other conceptual tweaks to keep it real), we have a useful SOA Semantic Web maturity model. To wit:


• Semantic Integration: These consist of all tools, runtime components, and services needed to retrieve, extract, and move semantic objects (i.e., data and metadata) from origin repositories; parse, validate, mediate, infer (deterministic and/or probabilistic) mappings among the semantic objects; transform the semantic objects; and deliver the semantic objects to target repositories, applications, services, users, and other consumers.

• Semantic Quality: These consist of all tools, runtime components and services needed to discover and profile source semantic objects; validate, mediate, de-duplicate, match, merge and cleanse those objects ; and enhance, enrich and augment it with additional, related objects.

• Semantic Repositories: These consist of all tools, runtime components, and services needed to organize, index, store, query and administer structured semantic objects; consolidate structured semantic objects into subject/topic-oriented, integrated, non-volatile and time-variant repositories under unified governance; and govern its controlled distribution to various target repositories, applications, services, users, and other consumers.

• Semantic Domain Models: These consist of all prebuilt master semantic governance objects (metadata, schemas, ontologies, glossaries, and vocabularies), plus semantic governance infrastructure that a semantic domain authority uses to administer the semantics of a particular process, platform, or other solution domain (e.g., MDM, data warehousing, enterprise content management, enterprise information integration, enterprise service bus, business intelligence) of a horizontal, vertical, B2B, organization-specific, regional, or other deployment scenario.

• Semantic Modeling and Mapping: These consist of all tools necessary to create business and technical definitions of master semantic domain models; discover, author, design, develop, index, query, visualize, browse, modify, version-control, access-control, import/export, and/or cross-reference one or more semantically distinct master data sets; and define and manage hierarchies, mappings and transformations among master semantic objects.

• Semantic Governance: This encompasses all repositories (metadata, ontology, policy etc.); collaboration environments (workflow, task management, exception handling, event-driven alerting, calendar-driven reminders, priority escalation etc.); controls (authentication, authorization, mapping/translation, version, validation, monitoring, auditing etc.); and other tools, components and services necessary to define, approve and administer domain models—including rules governing semantic integration, quality, and repositories--upon which semantic interoperability environments depend. This is sometimes known as a “semantic stewardship” environment.


Yeah...that seems about right....handy....I'm just about to launch into the Semantic Web vendor/product/market survey segment of my research for my upcoming BCR feature article on the topic.

I sorta feel I have a decent enough map of this jungle (though no map marks the quicksand that no doubt is everywhere).

More to come.