Friday, I wrote a piece
about my belief that we need to rename and reposition the Semantic Web
as The Graph Web. My position is that the Semantic Web is hard to
understand, and is poorly positioned in the minds of the public, and
also that the package of ideas that represent the Semantic Web is too
complicated as a starting point.
In Friday's piece, I made a statement that it seems made people believe I wanted to eliminate the idea of ontologies in the Semantic Web. Again, for those of you that are not experts in the Semantic Web, that would be most of you, ontologies are like a categorization system for the information in your database. It is a little like del.ici.ous tags but much much more rigorous and complicated. In any case the intent of my original piece was not to suggest we eliminate ontologies and so I want to clarify.
what I said was:
First, all the stuff that seems too difficult and not immediately beneficial must be stripped away from the core concept and message. This means, for the basic message, ontologies must go. We must stop talking about them because they just confuse people. And they are not needed to get the most important benefits.
The thing I was trying to get across was the "for the basic message"
part. The idea that I am proposing is a blend of marketing and product
marketing/positioning, perhaps with a little bit of product design
tweaking, though that may not be necessary.
The main point here is that by simplifying the basic message of what the Graph/Semantic Web is, it will spur adoption. The Wikipedia page for the semantic web is a mess. It is unreadable by mere mortals. And I suspect strongly that almost no one outside the tiny Semantic Web community knows what it is.
So my point is that by creating a *baseline* concept that is properly renamed to the Graph Web, that eliminates stuff that people can't understand in five minutes of conversation, we will be allowing the concepts promoted as "Semantic Web" the full opportunity to flourish. OWL, the semantic web ontology language, and ontologies need graph based data. Graph based data does not need OWL and ontologies. And so reformulating the baseline concept of the Semantic Web to being about graph data on the web is a much easier to understand initial concept.
This does not mean killing ontologies. What it does mean is that you present an initial package of ideas as a starting point of understanding that does not include ontologies except perhaps in some really trivial or simple way. Once people get the baseline concept they can graduate to a higher level of sophistication, which includes ontologies in a more substantive way.
Broadly speaking, the idea is that when you are creating new concepts, you can't force everyone to understand and accept all of them at the same time -- that is if you are looking for widespread adoption. For widespread adoption you need to create layers, or concentric rings, or whatever metaphor you want to use. You have to give people simple bite size chunks that can be usable and useful without understanding anything else. You can't take a first grader and start teaching them assembly language. You have to do it in phases.
In my last post I got a bit of feedback from people saying that we *need* ontologies to some things. And I don't deny that. Similarly, we need chemotherapy too, but only for people with cancer. It would not be prudent to suggest that everyone receive a course of such treatment just for good measure. On the other hand we do suggest everyone eat their green vegetables, brush their teeth, and exercise regularly. Some things are for the many, and some for the few.
It appears that James Hendler, father of the Semantic Web ontology system, is doing some very interesting work regarding using ontologies in very simple ways, which may fit nicely into my stripped down initial world view a baseline "Graph Web" view of the world. I do not know for sure whether I would consider it simplified enough, as I have not actually read his new book, Semantic Web for the Working Ontologist. But certainly, from talking to him, that sounds like the intent.
Right now, the positioning of the Semantic Web is not clear. The name is not clear. The concepts are not clear. The purpose is not clear. This is a shame because with a focus on the mindset and needs of the regular work-a-day developer, this stuff could be very appealing. And with layering, it can be easily understood. But right now the only people I know that are really jazzed about ontologies have a PhD, after their name. And so if the goal, as I think it should be, is ushering in the next wave of the web, call it 3.0, the excitement has to extend to people with a few less letters after their name.