Re:PhD student solves 2,500-year-old Sanskrit problem
by 082349872349872 on Thursday December 15, 2022 @08:41AM
(#33999329)
Close also counts in computation, where we routinely handle infinite (or at least combinatorially effectively so) numbers of possible inputs with a few general rules and a finite ("almost no", to use a technical term) number of exceptions.
[ Parent
| on HN ]
Re:PhD student solves 2,500-year-old Sanskrit problem
by asaph on Thursday December 15, 2022 @09:03AM
(#33999611)
Perhaps my impression that this grammatical problem should be treated with logical rigor, where even a single counter-example invalidates the solution, isn’t the way this is being treated. The way it was presented in the article was like it was a kind of logic puzzle, for which an elegant solution has finally been found. Sounds like it’s more like a better rule of thumb was discovered that reduced the error rate, not to zero.
Re:PhD student solves 2,500-year-old Sanskrit problem
by 082349872349872 on Thursday December 15, 2022 @09:17AM
(#33999811)
It's exactly a grammatical problem. And when we look at formal grammars on the computer side, we find in practice they consist of very regular expression and declaration sublanguages and an irregular statement sublanguage consisting of a finite number of exceptions.
On the language side, consider english verbs: there are a huge number of regular verbs, and a small number of irregular exceptions. Life be like that sometimes.