22 December 2007

I'm dreaming of RSS in => AIML out

I am still trying to get my head around the relationship between chatbots and the Semantic Web, or Web 3.0.... Any thoughts or comments on the precise nature of this relationship are welcome.

Converting from VKB back into AIML was my first crash course in working with XML dialects.... Since then the old lightbulb has gone off, or rather "on" I should say, and it suddenly dawned on me that the whole hullabaloo about Web 2.0 largely centers on the exchange of metadata, most often in the form of RSS, another XML dialect.

I was really stoked to learn of the work of Eric Freese, apparently processing logic using the Jena framework then manually(?) converting that RDF into AIML; however, I continue to wait for word of his "Semetag/AIMEE" example at http://www.semetag.com .

My understanding is that it is quite do-able, as in off the shelf, to pull RSS into a database and accumulate it there.... Could such a database of RSS not be used as a potential knowledgebase for a chatbot?

The missing element seems to be the processing, or DL Reasoner(?).... I have been unable to find any reference to such a web-based, modular DL Reasoner yet....

http://www.knoodl.com seems to be the closest thing to a "Web 2.0-style" collaborative ontology editor, which is fine for creating ontologies collectively, however falls short of meeting the processing requirement.

In short, I'm dreaming of RSS in => AIML out. At this point I would be happy with a "toy" or abbreviated system just to begin playing around with all this affordably (not least time-wise). So it seems what's still needed is a simple, plug and play "Web 2.0-style" (or is that "Web 3.0" style?) web-based DL Reasoner that accepts common OWL ontologies, then automagically goes from RDF into AIML....

3 comments:

Jean-Claude MORAND said...

I had the same dream in 2005 when I wrote the first edition of my book "RSS, Blogs: A new management tool" (pubished in French :<)). At that time I got the support from Dr. Wallace. I reached your blog tonigh, because I'm again considering this approach in another context. Indeed, Eric Freese work is a good start. RSS 2.0 is somehow limited... I guess we should dream an RSS 1.0 (based on RDF) firt and then I bet that in 5 years from now, DC and other standards will invade RSS and other applications will be possible.

mpalmerlee said...

Marcus-
How are you converting Verbots VKB to AIML, did you write a converter, or is there another program available? I know there is the AIML -> VKB converter, but I didn't know people were going the other way. In my option Verbots VKB's are much cleaner and easier than the AIML xml knowledgebases I've seen, although the AIML engine (and it's matching algorythm) is pretty elegant and powerful. Does it allow out of order matching do you know? Seems like if not that would be a neat addition to the AIML algorythm.

H_I said...

I was thinking of feeding my own bot with wikipedia pages, but yes! RSS feeds are a good option, too.

And I think it is easier with a learning chatterbot, than with AIML, where you have to translate the knowledge to a stimulus-response schema.
Am I wrong?