Is it time for the enterprise to lose the idea of the holy grail of a central repository of multi-sourced data?

Experts in the semantic web space think that’s the case. At the recent Semantic Web Technology and Business Conference in New York City, a few weighed in on the issue.

Take, for example, the take by David Wood, CTO of 3 Round Stones.  Why don’t we want to integrate data from various silos, as data warehouses do? “It always creates a project called Atlas, a company spends $30 million to do it, and it always fails because by the time it goes live it’s overcome by events,” Wood says. “We’ve been pretending for decades that we can centralize our data and we can’t and we should stop pretending.”

With a nod to Jeff Pollock, Wood discussed that the real holy grail is information interoperability. Our story here provides a look at how that’s facilitated in the update to Callimachus, a framework for data-driven applications based on Linked Data principles.

The theme of a central data repository was also echoed during presentations such as the one focused on Using Linked Semantic Data in Biomedical Research and Pharmaceuticals (see coverage of that panel here). “The idea that if we could get all this data in one place, our problems would be solved, turns out to be a ridiculous notion,” said Ted Slater, CTO at the OpenBEL (Biological Expression Language) Consortium at Selventa and also a consultant at Broad Reach Strategic Advertising LCL. The first open source release, Version 2.0.0 of the BEL Framework, lives at the OpenBEL Consortium. (Selventa, a biomarker discovery company that enables personalized healthcare, publicly released the language that represents causal relationships between entities in biology as represented in scientific literature and the associated BEL Framework as an open source platform earlier this year.)

“There’s a moral obligation almost to think about not where to put the information, but about how you will use it,” he continued. In 2012, in a world where the enterprise is no longer slave to slow microprocessors and a lack of disk space and memory that once limited choices of how and where information is stored, “we have to think about how we use this information. We have a lot more power.”

When it comes to enterprise data, said Yingxia Wang, senior manager at SunGard Global Services during a presentation on agile data analysis and reporting using semantic technologies, “converging into one data store… is simply not an option.”

It’s a change for many in enterprise IT, and it’s important to understand if there is some hesitancy about trying things a different way when it comes to the data for which they are responsible. A standard like R2RML  lets relational data sets be mapped to RDF and connected with the semantic model, for instance, but more seamless data interoperability may take some getting used to.

“You need to work with people who own these existing data systems,” says Lee Feigenbaum, vp of marketing at Cambridge Semantics, and consider their priorities. “They have to guarantee access to their systems, and the security of their data, because they can get audited next week.”