The Semantic Technology Conference 2007 [1] was an interesting event (even including the update on rule standards given by your erstwhile TIBCO rep). So what are semantic technologies, and what do they offer the CEP world today?
When someone mentions “semantics” in an IT context, people will automatically assume they are referring to “the semantic web”, a vision of the W3C’s leader Tim Berners-Lee that encompasses intelligent processing of distributed data sources from diverse web-attached devices. While “the semantic web” is not a very common phenonemon today, the technologies developed for it are increasingly being applied to more conventional software applications. The primary semantic web technologies from W3C are:
* RDF which is the semantic web data format, storing data in easily-manipulated tuples (actually triples [2]) allowing sophisticated graphs of information to be defined declaratively. Of course, if you decompose a database table into RDF you actually are losing information (the relational form of the data), so RDF is semantically poorer than current data persistence techniques – with the caveat that RDF-Schema adds back structure into RDF. Although RDF is usually thought of in the context of storing web information, there are a number of persistence mechanisms (aka databases) for storing large volumes of RDF in a non-web setting. The rationale for RDF is that tuples are more easily shared, require no maintenance when the (overlying?) schema changes, and can be manipulated more easily for example by ontologies…
* RDF-S is the schema for organizing RDF content.
* OWL which is the web ontology language, which defines relationships between RDF entities. So you could think of OWL as the class models for the RDF data, although that is a gross simplification; OWL relationships can be more sophisticated than UML type class models for example. OWL comes in a number of different versions (Full, DL, Lite) targeted at different audiences and which can also be incompatible with each other.
Other interesting semantic technologies from outside the W3C world include CommonLogic (often abbreviated to CL, is a formal logic language), and SWRL (albeit a W3C submission, is a rule language that extends inferences over OWL that came out of RuleML). Also there is a bridging standard that maps the W3C semantic standards (RDF, OWL, and also CL) to the conventional software world via OMG’s Ontology Definition Metamodel or ODM (in other words, the metamodels for the semantic web).
At the STC conference, there were a number of interesting presentations that touched on event processing and indicated where the semantics community is attempting to solve similar problems to the CEP community, but from a different angle: emphasizing the “meaning” and “knowledge” represented and updatable in CEP rules. We’ll review these later, and guess at the likely future direction for semantics in CEP.
Notes:
[1] Worth noting: this conference was well attended, and attendees were impressed by the organization. Congratulations therefore to the Wilshire Conferences team.
[2] Although RDF is about triples, many specialized RDF stores extend this to quads or use other representations that embed triples, for example to provide more efficient processing by adding index fields.
Disclaimer: the author is a contributor to the W3C Rule Interchange Format, which is considered by some to be a Semantic Web subgroup defining a semantic web rule language.
Credit: the author thanks Elisa Kendall of Sandpiper Software and co-author of ODM for advising on this post.
