The top part says a lot:
Input: a lot of sanitizer keeps covid away
Checking Pattern: Pattern 1421 "<thing1> keeps <thing2> away"
thing1 => a lot of sanitizer
thing2 => covid
Using pattern rewrite: When you have <thing1>, <thing2> will never come
a lot of sanitizer keeps covid away
=> When you have {a lot of sanitizer}, {covid} will never come
I'm unsure what the rest I didn't paste means, it doesn't seem to find more
than 3 matches and they are to me not a match, wtf....
Anyway, hmm, certainly with my plans my hierarchy/ heterarchy network would
learn Byte Pair Encoding parts and so it would activate the thing1 and thing2
nodes and any translates, including the node thing1 keeps thing2 away. As for
"When you have <thing1>, <thing2> will never come" this too can be stored in my
future net and can translate or entail the other thing1 thing2 sentence, and
then when I'd get the translated sentence "When you have <thing1>, <thing2>
will never come" my net likes to use words originally said in place of thing1
and thing2, achieving the same thing you try to achieve, but you're doing more
work hardcoding everything instead of treating all data parts as equal and
using BPEE, markov chains, word2vec, etc. This is no different from that, nor
them, nor a, nor home, nor frog, they are just words, bits n bytes, you don't
tell it which or what comes with/ is what, just the RELATIONSHIPS and
reoccurences.
------------------------------------------
Artificial General Intelligence List: AGI
Permalink:
https://agi.topicbox.com/groups/agi/T29d9a364b5df2085-M7ecd566ffb1626dbf37ac4ad
Delivery options: https://agi.topicbox.com/groups/agi/subscription
