That is an interesting paper. Unfortunately, as I tried to follow up with their
references I quickly discovered that the reference I landed on was written in a
technical (abstract) form that I could not interpret. Although I think
mathematics and formalizations expressed in mathematical terms are useful I do
not think that mathematical formalizations are going to solve NLP dilemmas
unless they are accompanied by and followed by a lot more mundane programming.
It is like they keep banging their heads on the problem and then they go off
and try to solve it with the methods that caused the problem in the first
place. Formal methods typically follow informal methods. In those rare moments,
like Cauchy's theory of limits, where the formal method does solve an
outstanding problem, it solves what we might call a narrow problem. That is not
to say that solutions to narrow problems never unleash new ways to analyze
other problems but once a bottleneck problem is solved it will take time for
developers to use that solution to *solve *all the other problems that the new
narrow method might be applied to. The use of filtering methods to build more
complicated *objects* of knowledge is a fundamental method of effective
compression for objects of knowledge.
------------------------------------------
Artificial General Intelligence List: AGI
Permalink:
https://agi.topicbox.com/groups/agi/T68be2fedf1f53ef2-M4f8d3c8d55907b01c3e92b72
Delivery options: https://agi.topicbox.com/groups/agi/subscription
