Semantic Technology's place in the enterprise; key, but low-key ?
Smaller than the more established Semantic Technology Conference in San Jose and nestling uncomfortably close both spatially and temporally to the more established European and International Semantic Web Conferences, ESTC nevertheless attracted some excellent presentations and an audience that seemed to have plenty to discuss.
With other events elsewhere in Europe earlier in the week my own visit to Vienna was brief, and focussed upon chairing the conference's closing panel discussion in the stunning surroundings of the Palais Niederösterreich; that ceiling had to be seen to be believed.
As I've mentioned before, the panel comprised six invited members and myself, and we'd been asked to discuss industrial uptake of semantic technologies.
We began with short opening statements from the panel, and some common themes very quickly emerged.
Peter Brown, for example, began by saying;
"Don't expect a warm welcome in the boardroom if you go in talking about ontologies, etc... Talk about information."
Judiciously applied Semantic Technology, he went on to say, should;
"hide the technology 'engine,' whilst ensuring that the user has a 'dashboard' and 'steering wheel' with which they can interact to drive the 'car.'"
Telefonica's Richard Benjamins was amongst those echoing Peter's sentiment, calling upon developers to;
"hide your technology."
Verizon's Michael Brodie and Vulcan's Mark Greaves shared similar insight from their roles evaluating the value of solutions pitched to their organisations for purchase or investment.
Michael Brodie suggested that, in his role evaluating third party solutions to requirements inside Verizon, he will;
"very rarely look at generic technologies... It is always about a problem to be solved, [rather than] about technology."
He went on to identify the metrics by which prospective solutions are evaluated, characterising them as;
- "legal issues;
- technical issues;
- product momentum [existing usage inside Verizon and elsewhere];
- standards [formal and informal, and including UI style, etc];
- strategic fit - is the problem being solved strategic to Verizon? Is the product being proposed strategically important to the organisation providing it [and will they therefore continue to support/develop it]?"
Vulcan's Mark Greaves joked that;
"We don't need to encourage industrial uptake of semantic technologies; I get more proposals than I have time to read."
On the basis of those proposals, he characterised three broad areas in which semantic technologies have a role; Business to Business [enterprise information], Business to Consumer and (most interestingly, he suggested) Consumer to Consumer. In each, he suggested that developers should;
"hitch on the back of existing [business/consumer] priorities"
In other words, we should be solving problems that people already know that they struggle with, rather than inventing new markets and new problems.
Mark continued;
"The cost of semantic data creation is tending toward zero. The quantity of semantic data is going toward infinity. The amount of 'bad data' is growing."
Better publishing and inferencing tools, he suggested, would make it more straightforward for the data being created to be 'good.'
BT's John Davies, the conference programme chair, talked about encouraging technology adoption inside his organisation, and commented that;
"eyes don't glaze over as much as they did twelve months ago."
John pointed to the visible successes of SearchMonkey, Calais, Powerset and others as important in helping to validate the messages he has been pushing internally for some time.
The European Commission's Marta Nagy-Rothengass talked about the Commission's ongoing investment in research, discussing the €300,000,000 ($430,000,000) currently allocated to knowledge engineering projects. She stressed the importance of these projects in ensuring European competitiveness, and in helping to make Europe 'the most developed Knowledge Society.'
Turning to discussion, the panel were asked for their views on the European Commission's traditionally generous funding of this area. I suggested that the project funding mentality, whilst good for research, might constrain true innovation and dilute the incentives to carry an idea to market.
Marta stressed the need to be constantly wary of 'over funding' making project staff 'lazy,' but also pointed to the European Commission's Competitiveness and Innovation Programme, which is specifically focussed upon carrying outcomes to market.
Peter asked if we were seeing;
"a generation of lazy [project-funded] fat cats, who never have to take anything to market?"
Richard disagreed, arguing that ready availability of project funding does not remove other opportunities to succeed. He pointed to the early funding of Semantic Web research in the United States under DARPA as one example.
Semantic Web Activity Lead at the W3C, Ivan Herman, joined in from the audience, noting the continued lack of good educational materials for those wishing to learn about semantic technology. He described a 'lack of expertise,' and noted how much harder it would be to secure European funding to develop a course in semantic technologies than to conduct some piece of new research.
Project 10X' Mills Davis countered that;
"market forces prevail."
He argued that techniques such as Desk Top Publishing were not taught by universities and colleges initially. These skills were learned on the job, as people needed them, and it was only later that courses grew up to cater for increasing demand.
Back on the platform, Mark stressed his view that;
"the Semantic Technology space is not underfunded in human capital"
He returned to earlier themes by going on to suggest that;
"startups in the US stand on the shoulders of the generosity of European project funding."
In closing remarks, many on the panel stressed the key importance of building a compelling (and complexity-disguising) user experience, and Michael wrapped up with the advice that;
"[A startup should] invest as much in understanding the market they're selling to as in developing the technology they sell."
Commenting after the event, Programme Chair John Davies said;
"I am very pleased that we achieved an increase in the number of attendees relative to last year. Also pleasing is the increase in the number of participants from industry as opposed to academia: up from 50% last year to 65% this year. My thoughts are now turning to next year's conference and how we can make it an even more useful forum for suppliers and users of semantic technology."
Next year's event will be from 30 September-2 October, and possible venues include Vienna, Berlin, and London. I know which would get my vote...
Image of the ceiling at the Palais Niederösterreich © 2007 Darijus Strasunskas, and used with permission. As more evidence of how small the world can be, this image was found via Flickr and was taken in a public building over a year ago, yet the photographer's web site begins his list of research interests with "semantic technologies, semantic interoperability" !
Image of the panel in session © 2008 STI International; the conference organisers. Available along with other event images here.