I haven't started programming this, as I need to think it out and do some other experiments, but the idea is it takes words' ATTRIBUTES, not the words, and tries to find the nearest meaning to it. Example: Take the sentence "Who are you?" Who=Question;Person, are=Question, you=Person;Me. It would then determine it was a question, about itself, so it would answer Who with Conversable AI, and are with I. An answer to that would be something like "I'm Conversable AI". --- Update Log V0.1.0 Assess question with vocab V0.1.1 Little bit better V0.5.0 (5/22/15) One word answer. Doesn't understand very well. V0.5.0 (5/22/15) Now can't answer with a subject. V0.5.6 (5/22/12) Minor Fixes
All by me, kinda based off Bayesian networks, but only kinda. --- Please only use current vocab (for the most part). --- Current Vocab: Hi You Bad What How Hello Hate Love
