I Have a (Semantic) Dream
Read Write Web, a site that I have in my list of favorites I respect, has published a list of the top semantic web products of 2008.
I have been looking forward to a successful semantic web application because the enterprise sector, where I think semantic technologies can really make the difference, needs a success in the consumer arena to move beyond its resistance to adopt these technologies. This resistance has been caused by two main reasons:
a) Too many companies with immature technologies have approached the sector in the last few years leaving behind unsuccessful implementations.
b) Too many companies marketed as semantics, technologies that are NOT actually semantics, promising performances that in reality were not reachable with their traditional approach, leaving once again behind unhappy customers.
With this in mind I was somehow disappointed with the Read Write Web selection because if these sites are the best 10, it looks to me like my dream of a real semantic web application is far from becoming reality.
What struck me first in the list is that there were names of applications that seemed they made the list because of the buzz and PR activities more than their functionalities or popularity. I am of course referring to Powerset and Hakia. While I took my hat off to the capability of Powerset management and VCs to negotiate an amazing deal with Microsoft, I think that in terms of application features, Powerset is still an incomplete product with limited performance, especially when compared to their initial objective. Things might change in the coming year given that their talented R&D team will have access to more resources. But for now is at least premature to consider them in the top 10.
Hakia has also raised a considerable amount of money with the same aggressive objective to become the new ‘natural language based’ Google. After having failed to gain significant traction, today they seem to have at least partially abandoned this idea and are trying to move to the enterprise. So, again, I am wondering where is the success that helped the application to make the top 10 list.
The list includes also a few applications that I love (like tripit.com), or that I think I could like if I had more time to surf the Internet (Boorah or Uptake) but that, unless I missed something, are not built on anything significantly semantic. In fact, Boorah and Uptake claim to use a natural language processing but most of the posts are extracted from sites that already have metadata, including the stars assigned by the user and the capability of understanding natural language by looking at the site, seem to be very keywordlike. Trip it, while very useful for road warriors like myself, instead doesn’t really use any semantic technology, as the information can be uploaded automatically in the site only if the emails sent have specific structures (this is why it works only for selected travel sites).
I want to add a few words on Open Calais. I am following with a lot of interest their effort and I am happy to see their initial success. I think that the limited semantic capabilities of their technology (which is based on the old Clearforest product) at the end will be a limit for Open Calais to enable real semantic web applications but I believe that the model is good and I would not be surprised to see some more complete semantic technologies following Calais’ steps.
I don’t really understand very well why the site has considered some of the remaining applications worth a mention (but I will use them in the next months to understand if they deserve a mention). In general, the applications that they listed are still niche applications, meaning that my dream of having a real, successful semantic application that demolishes the enterprise resistance to massive adoptions of these technologies is still a dream. Hopefully, I will see some of these in the 2009 Top Semantic application list.
Author: Luca Scagliarini