Professor James A. Hendler goes by the daunting title of 'Tetherless World Senior Constellation Professor' at Rensselaer Polytechnic Institute (RPI) in Troy, New York. Behind the title stands a man who has been closely involved with Artificial Intelligence (AI) research for many years, and someone recognised as amongst the progenitors of the Semantic Web ideal. Hendler is also Associate Director of the Web Science Research Initiative (WSRI), an activity that is being pushed hard by Sir Tim Berners-Lee (a Director) and others.
I spoke to Jim recently, and in a wide-ranging conversation we touched upon early hype around the promise of Artificial Intelligence, conflicting aspirations for the Semantic Web meme, and much more.
Early in the conversation, Jim highlights the importance of weakening the 'tethers' that tie us to computers, websites and applications; he illustrates the transition that we are already experiencing with the growing utility of mobile devices such as the iPhone, and points to the importance of metadata and structured information in underpinning the (virtual) connections (or 'tethers') between resources that will enable a commensurate loosening of the physical bonds that tie us to our computers and specific applications running on them.
Asked about the 2001 Semantic Web vision paper that Hendler wrote for Scientific American with Tim Berners-Lee and Ora Lassila, he agreed that reality had been slower than expected to follow the vision painted in that article. He disagreed with recent comments from co-author Berners-Lee, though, defending the future-gazing nature of the piece as both valuable and necessary.
Throughout the conversation, he returns frequently to the notion that advocates of new ideas (Artificial Intelligence, Web 2.0, the Semantic Web, and more) have a depressing tendency to communicate at cross-purposes, latching on to commonly used words and loading them with contradictory or confused meaning. Each time we returned to this important theme, my thoughts turned - irreverently - to Monty Python's Life of Brian, and the superficial yet divisive differences between the People's Front of Judea and the Judean People's Front.
Digging further into this idea, I asked Hendler (about 16 minutes in) whether he thought there was a 'shared vision' of the Semantic Web.
"I think there are several shared visions of the Semantic Web... We have people with very different ideas of what the Semantic Web is."
He goes on to suggest that the Semantic Web as it exists now falls into two main areas of utility;
- 'the heavy duty reasoning' or AI vision of the Semantic Web, based upon an extremely detailed and highly expressive model of a subject or domain, which is then used to analyse large bodies of data. Jim points to pharmaceutical companies as current beneficiaries of this vision, and the text analysis companies that have been featuring on this blog recently would also fall into this category.
- the data-driven vision of the Semantic Web, which is more lightweight, and geared toward the application of a far less structured world view to diverse data across the Web. RDF is more important to this vision than OWL, and it fits more readily with the existing Web
Hendler suggests that the former could be characterised as the Semantic Web and the latter as the Semantic Web, and considers himself to be more closely aligned with the second vision/ He stresses that both are important and valid.
"the move from talking about Chicken and Egg to talking about Chicken Farms."
In other words, there is enough data and enough technology available on the Web for us to move beyond research and development and into serious implementations.
Turning to the role of semantics in today's Web 2.0 applications, we touch upon arguments in a recent paper (pdf) that Hendler co-authored with Jennifer Golbeck. In the paper, similarities are recognised between the underlying principles of Moore's Law and the network effects we are seeing in data and connections between resources on the web today.
Hendler highlights the importance of context in our interactions with sites such as Flickr, and we explore the tension between increasingly vertical silos (where context can be inferred) and a desire to move more freely around the open Web (where context must often be declared).