Thursday, March 12, 2009

ISA's are the future of the web. Don't take our word for it! (You might want to take MIT's, though)

Technology Review, published by MIT, lists "Intelligent Software Assistants" among "10 Emerging Technologies of 2009". We couldn't agree more.

Check it out: http://www.technologyreview.com/specialreports/specialreport.aspx?id=37

(Notice how successful they were predicting the ones back in 2001)

Update: Here's the related press release

4 comments:

Selcuk Atli said...

I think there are two major flaws with the press release given in the link. First, according to the actual source, MIT does not endorse or mention Botego. However, the title of the article is misleading in the sense that it hints towards that direction. Second, semantic web is a term that is being used out of context too often. Semantic web means linked data, just like the world wide web is linked documents. It is the science and technology of integrating and reasoning over data from multiple sources. An intelligent chat bot service could be making use of semantic technologies, however I highly doubt that Botego has a semantic infrastructure.

Ekim Nazım Kaya said...

Selcuk, thank you for your comment. We have no intention to mislead readers, that's why we did not claim that "M.I.T. listed Botego" in their article. What we said is that they referred to the technology, which really is the case. They mention Siri in the article, and it is obvious that Siri is not the only company representing this technology, there are various others such as Numenta, Novomind, Novamente, etc. If M.I.T. considered Siri as the one and only developer of such technology, their title would include Siri, not virtual agents, so i have to disagree with you.

Regarding your doubts about our infrastructure, we respect your opinion, but we have no intention to discuss how we make use of semantic principles at this time. I would advise you to please keep following us, in case we decide to make this information publicly accessible.

Selcuk Atli said...

Ekim, thank you for your kind response. I would be happy to follow up with how you would be using linked data in your technologies.

However, the points I made was not about your blog post or the link you provided for the source under the link: http://www.technologyreview.com/specialreports/specialreport.aspx?id=37.

It is about the press release made under the link:
http://www.prweb.com/releases/intelligent/agents/prweb2254384.htm

The title of the press release is "M.I.T. Lists Intelligent Virtual Agents, a Semantic Web Application by Botégo, one of the Emerging Technologies of 2009". I am sure your intention is not to mislead, but I come to understand that the title is misleading, if you read the title out loud.

Furthermore, if you look at the original source, you do not see them mentioning semantic technologies on ISA's.

I'm doing semantic web researach at the tetherless lab at Rensselaer. There is a common misconception to relate to any new web technology as semantic. Linked data has a very high potential for changing the way that we interact with information in the future, but some of the work is at an experimental stage and there are issues of adaptability. I would recommend going through publications of Dr. Tim Berners Lee and Dr. Jim Hendler. There are some pretty good publications about adaptability and what linked data can do.

Again, I would be happy happy to see how and "if" Botego is making use of linked data in the future. Best of luck.

Selcuk Atli

Ekim Nazım Kaya said...

Dear Selcuk,

I believe perception is everything, so discussion on the press release is only meaningful up to a certain extend. Just be informed that if we intended to create a wrong perception, we would have been much more aggressive on the wording to make sure we achieve that goal.

Although we are already using linked data in every project to recognize synonyms, common typos and misspellings in the user input, we're using it in one of our projects in a way i think you imply. Please check this post out. Yazboz is a great database containing interrelated keywords that allows us to "interpret" the context in which a particular input resides. We're utilizing various similar tools most of which i may not disclose as i mentioned on my previous comment. If you check out the following posts along with Yazboz website, you might realize there's no doubt that we "DO" use linked data:

"Bring me fifty fantastic transcripts!"

You can't think outside "our box"