What’s on the horizon for the semantic web? It was a question pondered by expert panelists last week at the Semantic Technology and Business Conference in San Francisco.
Siri – and its possible clones and descendents – came up as a signpost on the road ahead, pushing the notion of personal assistance ever forward. “Siri made a huge splash. It opened eyes to the idea of not just using semantics to move information around but using a natural language system with some semantic interpretation to perform actions, like putting reminders on your phone,” noted Mark Greaves, director of knowledge systems at Vulcan.
Speaking of phones, it’s the idea of context-aware personalized mobile assistance that has Deborah McGuinness’ phone ringing off the hook with questions about the trend, including from leading vendors in the mobile chip and device space – a good sign that the topic has juice.
The Tetherless World Chaired Constellation Professor at Rensselaer Polytechnic Institute and CEO of McGuinness Associates noted that the semantic wine and food pairing app she created a few years back (see our story here), WineAgent, has gotten more attention in the last year as its features set incorporated mobile platform capabilities such as global positioning information and input from social tools such as Facebook. The reason, she thinks, “is that contextualization made it more interesting to the ‘normal’ public. So that’s one indicator of where this comes out in the real world.”
The art of personalized assistance by way of semantics has not as yet been perfected, of course, but at least it’s trending in the right direction. “Most people would say Siri is mostly a success,” McGuinness said. A lot of people can’t get what the Siri commercials show to work quite as seamlessly at home as it does on TV, she noted, so there’s a ways to go. While, “as an AI professor [I know] there’s a whole lot more that Siri and things like it can do, …it’s a viable, workable example that shows a vision of where things can go.”
Also on the Horizon
Steve Harris, CTO of UK-based Garlik (who also gave an opening keynote address at SemTech), now a part of Experian, discussed the SPARQL 1.1 Query Language for RDF data. The Last Call review period of the W3C Working Draft specifications just concluded earlier this month.
“Exciting things are coming,” he promised, “things you can’t do with relational databases. People are always saying why can’t we do certain stuff in SPARQL but the capabilities you get now are quite impressive.” That includes property paths to query arbitrary length paths of a graph via a regular-expression-like syntax, though how optimized has been another matter, he noted. (The semantics of property paths has been a subject of debate over performance in the community and led to adoption of an alternative design for property paths by the Working Group, as noted here).
The panelists also considered usability and accessibility related issues for making semantic solutions more palatable to the general public, especially in the enterprise. Bill Andersen, vp of R&D of Highfleet, which develops semantic-driven database systems, pointed out that one thing that customers want most is getting new value from existing data – and that they want “a pretty picture” to achieve their ends.
“But we are a back- end company, which is why we teamed with fluid Operations,” which provides a flexible and extensible widget-based user interface for integrating and connecting data from different sources. But finding fluidOps was serendipitous, he said, and “if you’re looking for some method by which these synergies can come together, I don’t know what that is.”
Greaves made the comment that increasingly semantic technologies are being provided as services on the web, along the lines of efforts like Open Calais and sameas.org. And, in the next couple of years, expect more popularization and democratization of semantic offerings. “I think,” he said, “that we’ll see people putting out more personal and atomic kinds of semantics via things they use like Facebook.”