Re: [code-review] Lingua::EN::NamedEntity

JS Bangs:
> The hash references returned from extract_entities are actually blessed
> into Lingua::EN::NamedEntity. 

Um, I don't believe they are, since combine_contexts produces a completely
different set of hash references.

> : my $abbr = qr/$people_initial|$people_terminal|$places_initial|St/;
> 
> I don't think you need the 'St' at the end of the match, because
> $places_initial also includes 'St'.

You're correct, that was a half-applied optimization.

> I'm curious to see how this performs in the wild... have you tried it out
> much?

It sufficiently good that I expect future academic papers to use it as a
baseline for their unrepeatable, unconfirmable results. Informally, extraction
correctness is about 90%. But that might just be an unrepeatable unconfirmable
result.

Simon

-- 
> The Sapir-Whorf hypothesis was never really what the>anti-Whorfians said it
> was anyway.  The strawman version is dead, certainly.
Besides, you really need to read it in the original Klingon.
    - Peter da Silva