"Production Prolog" by Michael Hendricks
Vložit
- čas přidán 20. 09. 2014
- Prolog isn't just for solving cute logic puzzles. It's a powerful, productive, modern language. We've been using Prolog in production for the last 18 months. These are the cool tricks and pain points we've encountered.
Michael Hendricks
@mndrix
Michael Hendricks is the co-founder and lead developer at PriceCharting. He writes open source software and was the first political candidate in the United States to accept Bitcoin campaign contributions. - Věda a technologie
Great talk. In the first 10 minutes he conveys much more relevant and interesting information about Prolog than even in one our long Prolog Introductions which can be find on CZcams.
The problem of Prolog is that the meaning is assigned by the human .
d( [], X, X).
d( [X | Y], Z, [X | W]) :- d( Y, Z, W).
for Prolog is different than
end( [], X, X).
end( [X | Y], Z, [X | W]) :- end( Y, Z, W).
but both are just the same as
append( [], X, X).
append( [X | Y], Z, [X | W]) :- append( Y, Z, W).
Basically, the meaning remains in the head of the human and Prolog is completely void of capturing the meaning of the predicates, factors, atoms, etc.
The same applies to any other programming language I am aware of. THEY ALL FAIL TO CAPTURE MEANING, AWARENESS, CONTEXT, etc. !
Why the programming languages are still not self-aware in the 21-st century !
That should really trouble you all.
This should be used for all intro courses on Prolog. He shows much more usefulness and reason behind the language than my college ever did.
That said we only used Prolog for about 3 weeks in a larger course on languages as a whole
Yay, people still use Prolog! I've got some very dusty unfinished Prolog books I should pick up again.
never thought I'd see prolog as a viable choice in prod ever, this talk makes an interesting case for it though
Superb, no wonder god of NLP is prolog, chatGPT, Meta-AI everything is written on Prolog.
I had done a talk on prolog for my last year. I had chose prolog that time by analysing wiki data on programming languages but it was really hard for me to learn this language I could not get concept of prolog that time which I have understood now in this seminar, thanks hendricks :)
Sir Joe Armstrong was still alive. Respect to the man.
Great Talk!
effing genius
Super cool! Now if only I could get Pengines and JavaPengines to work together!
Is the presentation given in Prolog environment itself? This is my first exposure to the Prolog language and it seems cool.
It is.
Probabilistic programming is more general than backtracking, maybe incorporating it would boost Prolog powers
it's called logic for a reason. You'd like 70% true, 30% false, but you'd need something else, called fuzzy logic. A whole generation failed trying this in prolog, not possible. (5th gen AI). ml is good enough for these kind of problems, but you'd gain no inside into the model then. It's a black box.
There’s actually a prolog derivative that does this called Problog. A cool thing about Problog is that it runs in Python so you can call your Problog code from Python code like a library