Earlier this week, Gigaom announced their latest event; Structure Intelligence.
“In the past year we’ve seen massive growth in Artificial Intelligence (AI) and deep learning. Our own Derrick Harris has been covering this area for years and we have decided it’s time to give this rapidly growing area a platform (and conference) of its own.”
Personally, it’s great to see this area getting real attention from Gigaom. But I’m reminded of just how long some of these issues have intersected (or, often, passed close-by) my own career…
Back in the mid-1990’s, I was researching a Ph.D in the Archaeology Department at the University of York. Nothing particularly semantic or artificially intelligent about that… except that a fellow student was building a neural network that inferred all sorts of things about Viking sheep, based on cues derived from photographs of their excavated teeth. And expert systems were in vogue, for a while, doggedly following rules to identify such esoterica as the precise Dragendorff form of Samian pottery fragments. I stuck to teaching computers how to build maps of the archaeology below our cities. Much prettier.
Fast forward around ten years to Talis where, for a while, a team was assembled that pushed the envelope on what might be possible with an open, extensible and connected semantic platform. Alongside the technical developments, significant effort was invested in raising awareness of the potential around semantic technologies. Podcasts, two blogs, a print magazine (!), invitations to present globally. All now, sadly, gone from the Interweb as the company turned in a very different (but, given the unanticipated arrival of a global financial crisis, possibly understandably safe) direction.
As Richard Macmanus noted at the time, there was a brief flurry of interest in ‘semantic’ amongst the Web 2.0 crowd…
There was also a far deeper, and more sustained, interest in putting core semantic technologies to work in solving big, hard, slow problems. Events like the long-running Semantic Technology Conference (coming up in San Jose this August) offered a rare opportunity for academics and business leaders (and lots of spooks) to mingle. Other events tended to cater to a more academic crowd, making it harder for the big ideas to really get tested in business situations.
ZDNet had a semantic web blog, which I wrote for a number of years. Subsequent site reorganisations have broken a number of the URL redirects, but all the content does still seem to be there if you craft the right searches. There was also a site called SemanticWeb.com, now part of the same organisation as the Semantic Technology conference, where I wrote a regular column for a while. There, too, not every link survived the site’s transition unscathed.
Here in Europe, the funding stream that’s now concerned with ‘big data’ and the ‘data value chain’ was far more effusively intrigued by semantics, natural language processing, pattern recognition, and related matters. I’ve reviewed a lot of proposals there, and the change in the language of funding calls and bids has been interesting to observe.
And now, as Gigaom’s post noted, “we’ve seen a massive growth in Artificial Intelligence.” Everyone seems to be talking about it, and they don’t necessarily always mean what an AI researcher would mean. Much of what we’re seeing goes back to those enthusiastic days before the crash, even if some of the language and emphasis has shifted. Bottlenose, which I wrote about this week? Brain-child of Nova Spivack, of Twine fame in 2008. Microsoft’s Machine Learning, Business Intelligence and Analytics stuff? A lot of it is underpinned by people, technology and ideas from Powerset. Google’s Knowledge Graph, most recently seen telling us what the symptoms of a common cold might be? There’s lots of Metaweb and Freebase under the hood.
The language has definitely shifted. Artificial Intelligence, which was stuffy, esoteric, academic, and a bit pie-in-the-sky even to the semantic technology crowd, is now the term du jour. Semantic Web is only mentioned in whispers, if at all, although some of it sees practical and pragmatic adoption as part of initiatives like the Google-backed schema.org.
I made a conscious decision, when I left Talis, to focus on the emerging opportunities around cloud and data. Hence the name. Cloud, increasingly, is table stakes. It’s background, foundation, platform. It’s not interesting for itself, but for what it lets you do. And what it lets you do is gather, process and act upon oodles and oodles of data. Software like Hadoop, and data-based trends like open data mandates from Government create a fertile mix of capability and content. Computing is faster, cheaper, better. Data is better, more comprehensive and more available.
Now, maybe, all that semantic stuff can have its day… powered by cloud and data. And Structure Intelligence can play its part in telling — and shaping — this next chapter.
Disclaimer: I am an active member of Gigaom Research’s Analyst Network. They pay me for various things. They do not pay me to plug their events.
Image is a close-up of the modern replica of Charles Babbage’s Difference Engine, photographed by Larry Johnson.