Callimachus Update Has Big Implications for Data Interoperability

Callimachus is getting an update. It’s been quiet for a few months over at the framework for data-driven applications based on Linked Data principles, but with good reason, says David Wood, CTO of Callimachus project sponsor 3 Round Stones. 3 Round Stones also offers Callimachus Enterprise, winner of this year’s Startup Competition at the Semantic Technology and Business Conference in San Francisco.

Big things were on the menu for this release, which should emerge from beta today. To date, all the RDF that Callimachus has dealt with has been local to it, Wood explains. “People have been saying for ages, ‘But I don’t want to copy the LOD cloud into Callimachus to deal with it. I want to deal with a lot of data out there in the world, in enterprise systems, an Oracle server, or the LOD cloud,” he says.

The new release takes on the challenge of dealing with data that’s external to Callimachus.

“We probably rewrote half of the internals of Callimachus in this release,” Wood says. “Two big components are an implementation of the persistent URL concept (PURL), and XML pipelining, an XPROC standard from the W3C,” which it uses to gather documents from the web, and convert them to RDF. And, once that’s done, combine them on the fly, presenting them via the Callimachus template engine in human- or machine- readable formats. The PURL work involved the creation of a new type of PURL, called Active Purl, that actively gathers, creates and displays all results, as the mechanism to implement distributed federated queries. “It’s not just redirecting to some other source but actively participating in the creation of data it returns,” he says.

The PURL service implementation helps companies that want a single URI they can resolve on the web to serve up everything the company knows about a concept – a disease, for example, for a pharma firm, whether that disease is referred to as a heart attack or myocardial infarction. With the PURL service implementation, clicking on a link to retrieve all known information about ‘heart attack’ might get some data from an Oracle database, from Sharepoint, from the LOD cloud, or WebMD. “Each of those results comes back in different formats,” says Wood. “It’s relatively easy to get that data back into a format like XML or plain text or something you can work with. By passing results through XPROC pipeline, you can scan and convert each individual data response to RDF.”

“It’s a big deal because for the first time Callimachus cannot only serve as a Linked Data management system for local data, but can gather data from outside and still present it live,” he says. “Inside the enterprise firewall is a good use of this functionality. There are different formats, different proprietary systems that never were designed to work together. It provides a way to let those systems interoperate that is cheap, lightweight and sits on top of existing infrastructure.”

What begins life in the open source Callimachus project ultimately makes its way into the Enterprise implementation, which offers capabilities such as a more scalable RDF database, single sign-on and commercial support.  The Enterprise build should follow the open source release a month or two later.