| I've been trying to drag my focus back around to more AI / Machine Learning stuff for a while, so to give myself a sandbox to play with (and a starting point) I built[1] an AIML[2] bot using program-ab[3]. Then I wired it up to XMPP[4] so I could chat with it. Then I turned it into an OSGI[5] bundle so it could run in Apache Felix[6]. And I installed ejabberd[7] on the fogbeam.org domain so I would have a convenient place to play with bots that talk XMPP. Right now it doesn't do much besides replying to a few stock inputs like "hi" and "howdy" and "Hello", but I just started on this about a week or two ago. I also started adding "@ commands", and it will respond to "@time" with the current time. From here on out my plan is to get back into studying AI techniques heavily again and see what things I can do to make this bot "smarter". I have done a lot of work based on Semantic Web technologies, so I'll probably start working on how to do some stuff with this based on using an RDF[8] based knowledge-store. There are a couple of OSS projects out there for translating natural language queries into SPARQL[9], so I might soon try wiring this up to where it can use dbPedia and / or Wikidata, as well as other Linked Data[10] sources to answer questions. I'm not strictly interested in any Turing Test or Loebner Prize[11] stuff, as I'm more interested in making something useful than something that emphasizes "tricking" somebody into thinking it's human. That said, if it ever worked well enough, I think it would be fun to enter something like the Loebner contest, but that's not the main goal here. Also, right now I'm reading Hofstadter's book Fluid Concepts and Creative Analogies, and have been playing around with ACT-R[12], a popular "cognitive architecture" for doing AI research. [1]: https://github.com/mindcrime/LearningAIML [2]: http://www.alicebot.org/aiml.html [3]: https://code.google.com/p/program-ab/ [4]: http://xmpp.org/ [5]: http://www.osgi.org/ [6]: http://felix.apache.org/ [7]: http://ejabberd.org [8]: http://www.w3.org/RDF/ [9]: http://www.w3.org/TR/rdf-sparql-query/ [10]: http://linkeddata.org/ [11]: http://www.loebner.net/Prizef/loebner-prize.html [12]: http://act-r.psy.cmu.edu/ |