Amir Plivatsky

It seems this is an inconsistency in Debian testing's installed packages, as the autoconf macro for Python detection tries to use a depreciated package that is not installed. But since...

I also use this exact `aspell` version in my system, and it also returns "seasoned" first: ``` text $ aspell -a LG (master $% u=) @(#) International Ispell Version 3.1.20...

As you can see, on my system `aspell` also returns "seasonal". Using `link-parser -v=9 -debug=guess_misspelled_word` reveals it is not being used (on my system): ``` text linkparser> the seasand lakes...

My intention with spell-correction was that it would consider all the run-on splits and the first N single-word corrections (`!spell=N`). This is based on the assumption that the spell-correction library...

> > This is based on the assumption that the spell-correction library gives the best matches first. > > I don't think that's right. I think it provides matches with...

``` text linkparser> the seasand lakes are hot Trace: guess_misspelled_word: spellcheck_suggest for seasand: - sea sand - sea-sand - seas and - seas-anda similar - seasoned ``` It turned out...

> The combinatorial explosions that bother me the most, these days, are from the auto-generated dictionaries. These are unusual in several ways (depending on which style I work with) In...

This issue contains several ideas that may be separated into new issues (or maybe discussions?): 1. Reduce combinatorial explosion by selective extraction from the parse set (`struct extractor_s`). 2. Handle...

I produced the disjuncts of all the unique dict expressions of words. There are ~170M disjuncts in this disjunct list. Many of them don't make sense for me. For example:...

> There's a combinatoric explosion, there's no particularly obvious way to limit it. Maybe there is a need to make more efforts in pruning. I thought of more complex algos,...