In the final installment of this two-part series, Dean Allemang and Scott Henninger look into how the insights about semantic web education have an impact on the adoption of semantic web technologies.
When we speak of ‘technology adoption’, we are referring to the uptake of a new technology by a large group of people. For technologies like cell phones and DVD players, we are speaking pretty much of the general population, cutting across age groups, gender lines, and professions. For other technologies, like blood pressure monitors or language compilers, we have a very specialized target group in mind. But regardless of the technology audience, technology adoption is capricious, relying not only on technical superiority but on a number of issues like marketing, timing and image.
Technology adoption often incites highly emotional responses, especially when technologies are competing. The fierce ‘format wars’ in the 1970’s and 1980’s between Beta and VHS still serve as an example of technology marketing. Bring up the topic of Mac vs PC in almost any group of people and you are sure to have a heated discussion on your hands. Mobile telephony is a particularly bloody battlefield; in the most recent volley, Droid has launched an assault on the dominance of the iPhone. Web technologies are no different; the syndication formats RSS and Atom were the subject of a bitter format war starting around the year 2000.
The Semantic Web standards – RDF, RDFS, OWL, SPARQL, and SKOS are new technologies whose proponents hope will be adopted on a wide scale. In order for many of the Semantic Web scenarios speculated by the W3C to come to pass, these technologies have to be adopted on a large scale. But like any new technology, they face fierce competition from current technologies.
While technology adoption isn’t an exact science, there are some principles that can, other things being equal, clear the path. In the context of Semantic Web Technology, we’ll talk about just three of them in this article: Simplicity, Education and Image.
Simplicity
Whether a technology is to be adopted by a mass market (like telephones and video recorders) or by developers (like RSS or HTML), simplicity is an advantage for adoption. A simpler technology is easier to explain, easier to get started with, and easier to manage. This is often at odds with a technology’s power; a more complex technology (like SGML vs HTML) is usually much more capable. But for adoption, simplicity usually dominates.
This presents a real challenge for innovative technology. An innovative technology brings something new to the table but has to be simple to be adopted. For the Semantic Web standards, the W3C has chosen to address this by layering the standards (as shown in the famous “layer cake” in figure 1). While the whole stack is quite complex, the complexity comes in one layer at a time.
Figure 1. Semantic Web “layer cake” diagram from the W3C
In the case of a computer language (taken in the general sense, so that RDF, RDFS, OWL, SKOS and SPARQL are all computer languages), simplicity can also refer to the number of features in the language. The strategy of the SPARQL team was to publish a very simple version 1.0 of the language, and add more features later on. While this made for a frustrating version 1.0, it also made for a version of the standard that could be evaluated early and easily. This has made SPARQL adoption much easier than, say, adoption of a rules language. SKOS took a similar strategy.
The OWL committee chose the layered approach – in OWL 1.0, there were three layers, including a simplified “OWL Lite”. OWL 2.0 has filled in a rather large array of features – it remains to be seen how well this will far for widespread adoption.
Appearances can be as important as substance. In order to be adopted easily, a technology not only has to be simple, it has to look simple. RDF is a tremendously simply standard – but it struggles to this day with a needlessly complex XML serialization. Especially for XSLT programmers, RDF appears to be very complex indeed. This appearance continues to hamper RDF’s adoption.
Education
Education is a key feature of adoption – an industry can’t adopt a technology if the practitioners can’t learn to use it. For mass market industries, this comes down to good user experience design. Can your mother use her cell phone? Can you program your VCR? Ease of education is a driving force in the design of new technology. Things that are easy to learn are easier to adopt. Those who live on the Mac side of the PC war cite this as a major advantage of their player.
This poses a challenge to information technologies like the Semantic Web. On the one hand, a representation system like RDF needs to have a strong formal basis, so that it is possible to verify that an implementation does what it is supposed to do. Relational databases have the relational algebra, and RDF has a strict logical formalism that defines exactly what a triple really means.
The problem shows up in education – while formal logic may well have very broad applicability in many fields, it is not a topic that many people – even professionals – have a lot of training in. In fact, formal logic is the part of mathematics that is most likely to instill a sort of ‘math-phobia.’ If a technology requires that its users learn logic in order to understand it, it will have a severe educational barrier to adoption.
This is a special case of a common fallacy about technology adoption. If those who will adopt the technology are themselves technologists (e.g., programmers, database architects, developers), then, so the fallacy goes, we can expect them to learn whatever technical details are necessary. But technologists are actually people, and have limited time to spend learning new things. Even when marketing to technologists, education costs are important.
Just as it is possible to use a database, and even to query it, without understanding the details of the relational algebra, it is possible to use RDF without knowing anything about its formal logical semantics. This alleviates the problem for RDF. For OWL, the situation is much more challenging. The basic notion behind OWL is fundamentally logical – inferencing in OWL follows a particular logic. But if one first needs to understand the nuances of formal logic before they can use OWL, then there will be a serious problem when it comes to technology adoption. The education barrier of teaching the world logic is too severe a barrier.
In the case of OWL 1.0, it is possible (with some effort) to delay the introduction of logic until quite late in the description of the language. In _Semantic Web for the Working Ontologist_, we don’t introduce logic until chapter 13, long after RDF and OWL have been discussed. So it is possible to keep logic at ‘arm’s length’, at least for a while.
Image
A strong influence on technology adoption is image – how a technology is perceived can be more important than any real technical features. A primary concern during adoption is whether it is ‘ready for prime time.’ Most fundamental technologies, like the Semantic Web technologies, begin with research projects in academic institutions. The transition into ‘prime time’ readiness is associated with getting the technology out of the labs and into industry. The polite word for a technology that is not ready for adoption is ‘academic’; if a technology is seen to be academic, then it hasn’t made the leap from university to industry.
The Semantic Web has suffered for a long time with the academic label. With its (apparent) emphasis on formal logic, still considered by many to be a purely academic topic, the Semantic Web continues to meet adoption resistance on the grounds that it is not industrially relevant. While the work of professional academics on the W3C Semantic Web committees is laudable, the high level of academic participation only contributes to the image problem.
Many efforts are underway to alleviate this issue. The World Wide Web consortium has collected case studies for successful adoption of semantic web technology. The International Semantic Web Conference (historically a largely academic conference) instituted the Semantic Web Challenge, to highlight industrial successes of Semantic Web technology. The recent announcements by Google and Yahoo of their efforts with RDFa have done a lot to dispel the image of RDF as an academic project though this hasn’t helped a lot with perceptions of OWL.
Conclusions
Developers of a new technology naturally want to develop a high-quality product. And as their understanding of an area – like the Semantic Web – grows, so does their ability to produce high-quality technology. But the success of a new technology depends on a multitude of factors beyond its technical quality. Thus, from an adoption point of view, technological soundness has to be traded off with other facts, like education, simplicity and achieving an appropriate industrial image.
While the W3C and other proponents of the Semantic Web have done a lot of work to achieve this balance, it is too early to claim victory. Semantic Web standards have gained some traction in high-profile places, but many barriers remain. The barriers come from general awareness (“what can it do for me?”), as well as from developers accustomed to other, more conventional, technologies. The image of the Semantic Web as a an academic exercise of interest only to logicians is relenting, but not gone. Through the years, Semantic Web industry insiders have put in a lot of effort to promote the adoption of Semantic Web standards. From training programs that take students from introductory levels to advanced modeling, to books like Semantic Web for the Working Ontologist, which strives to remove formal prerequisites for understanding the Semantic Web. It is our hope that together, these efforts will help the Semantic Web achieve broad adoption in a wide range of industries.
References / Resources:
W3C Semantic Web Used Cases http://www.w3.org/2001/sw/sweo/public/UseCases/
Semantic Web for the Working Ontologist http://workingontology.org/
Comment on Research and Adoption http://xkcd.com/678/
Semantic Web Challenge Winners http://iswc2009.semanticweb.org/wiki/index.php/ISWC_2009_Awards#Open_Track
Google and RDFa http://radar.oreilly.com/2009/05/google-announces-support-for-m.html
Yahoo! and RDFa http://rdfa.info/2008/03/14/yahoo-into-semantic-web/
SPIN API: http://spinrdf.org/
TopQuadrant Training Program http://topquadrant.com/training/training_overview.html
Comments
Education – Semantic Technology Training
Love this blog, thanks! I’m on my way to spread it around already…
For anyone looking for EU trainings in Semantic Technology: www.tenforce.com/training