I have been running through a few things and thinking based off of a few things I have read..
a) I have abandoned the 'neural' path mimic technique that many people seem to like. I disagree with this mainly due to the fact that I have read in the human brain all information is retrieved within 4 'synapse?'..but I cannot think of a way for data to be categorized in this way with an order.
b) From reason a, if there is no order and it is seemingly random as to where each chunk of data is stored that makes it pretty difficult to find the data within 4 fires. (that could become extraordinarily slow after having 'taught' the ai thousands of nodes worth of data).
so back to point a, this means that there needs to be some sort of order to the pseudo thought processes an ai must use to retreive and store data. thus, i will accept data must be stored in much longer narrower trees. rather than wide shallow trees. my thoughts here are now, how to make the ai learn where to retreive/store data. this is all mostly speculation because i am having trouble finding online resources for non-robotic applications. the second of my trouble is word cognition. i wrote down 3 sentences that i would try to figure out from a computational stand point.
very abstractly the last sentence could be construed to mean both the fruit and the amount of shades of similar color. has anyone else ever tried doing this? the fact that the last sentence says 'coloring box' implies orange as a color, but you could say just 'box' and have been previously been talking about coloring or gesturing toward a box of crayons. i have found a single source geared towards solving this sort of thing but it does not handle prepositions nor does it handle actually translating them - pretty much all it did was what i did above. my only thought is to let the ai ask a question if the meanings do not become clear.
I think much of this is worth discussing, particularly the latter.