Hacker News new | past | comments | ask | show | jobs | submit login

Thankyou for the illuminating explanation.

Do you know if there is a way to encode these reasoning chains so as to assess their validity automatically? Like a knowledge graph with propositions linked in some way?

I have been reading about knowledge graphs but haven't seen this kind of application yet, but I think it should exist.




It depends on the (and I will use the term loosely here) "category" of interest. ie which kinds of implications you are talking about. Perhaps unsurprisingly, for mathematical propositions (constructive), these things do exist, and almost certainly surprisingly, they tend to be implemented as libraries of functions over a particular type system (constituting a less loose use of the word category).

For more statistical propositions, such as those that occur in science. Those exist but they're very flawed, and it's a tremendous effort to get anything useful out of them, on account of them dying when we try to look inside. If you haven't caught on, I'm talking about humans. We have yet to replicate such things automatically. You're certainly welcome to try though :).


>way to encode these reasoning chains

Symbolic logic. Learn: Truth tables -> rules of implication and replacement -> first order predicate logic -> quantifiers -> second order predicate logic.

Chapter 5 or 6 through 8 of Hurley's intro to logic does a great job imo.


This is the field of study called Logic. For a good page with resources see:

https://www.logicmatters.net/




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: