[claude2]

Another source, apparently, a very powerful chatbot, is the equivalent of a strong intelligent knowledge, by DataBank, WIKI. Must register for an account, and the source code installation, this project is free, in 2012.

http://wiki.opencog.org/w/The_Open_Cognition_Project

The chatbot is a demo of the OpenCog natural-language processing pipeline; it is NOT -- repeat NOT -- a demo of OpenCog reasoning, deduction, or anything like that. The chatbot is as dumb as a rock, as thick as a brick; it doesn't really do any "reasoning".

It can, however, answer simple English-language questions. It does so by several methods. The most straight-forward is to compare the syntactical structure of the question to previously entered sentences. Thus, for example, if the chatbot is told that "John threw a rock", and then asked "Who threw a rock?", direct comparison allows it to equate "Who" -> "John". These comparisons are done on the syntactic structure (dependency parse) of the sentence: that is, the parsed form of the sentences and questions are compared. This allows for some sophistication: the system can correctly answer "What did John throw?" for the above input, because dependency parsing will correctly identify "throw" as the head verb of the sentence. An additional layer of abstraction, to pick over and normalize prepositional phrases, is described in the opencog/nlp/triples/README file.

[squarebear]

Interesting but I would have thought most bots worth mentioning could already handle "John threw a rock. What did John throw?" type questions.

[Art]

Interesting but I would have thought most bots worth mentioning could already handle "John threw a rock. What did John throw?" type questions.

Steve, what type of bot(s) would these be? Pandora or similar AIML based bots?

[squarebear]

This sort of question style is popular in the Loebner Prize and so I would assume any bot who wishes to take part in that, would be able to handle this style of statement and follow up question. I can't speak for other languages but yes, AIML is quite capable of doing this.

[claude2]

1) Squarebear, Art, Interesting but I would have thought most bots worth mentioning could already handle "John threw a rock. What did John throw?" type questions.

I think this is a comparison of the author. For my part, I know Hal, 'my Hal5' Could distinguish, on an issue ', and' Reply '. Nothing is comparable to Hal. But the author has certainly done the analysis on the fluidity, the intelligence of a 'Bot'.

2) Squarebear, Art, This sort of question style is popular in the Loebner Prize and so I would assume any bot who wishes to take part in that, would be able to handle this style of statement and follow up question. I can't speak for other languages but yes, AIML is quite capable of doing this.

I have 7 different for Hal5 brain. A brain that has the possibility, of stored, a specific subject, and himself, back on topic, with other matter. Hal, this possibility has a very long time in advance, on the chatbot, Hal is able to make a real dialogue, since I know Hal, I learned a lot about artificial intelligence. Thanks to you all, and also thanks zabaware forum. But technology, currently, for AI, most applications are iPad. For other situations, some companies use very means to push for the adaptation, or rehabilitation of problems, or the handicapped, the human speech.

Thank you for your comments my friends.