cheerful parsers

[trochee]

[from work] I've just started working with the Dan Bikel parser implementation that subsumes the Collins parser.

It's an effective, clean parser that is written in Java. It's nice and configurable, and a good piece of software engineering. But my favorite bit so far is the training output.

It ends (emphasis mine):
...
Writing out canonicalEvents...done (00:00.000).
Writing out lexPriorModel...done (00:01.152).
Writing out nonterminalPriorModel...done (00:01.237).
Writing out topNonterminalModel...done (00:00.007).
Writing out topLexModel...done (00:00.254).
Writing out headModel...done (00:03.404).
Writing out gapModel...done (00:00.008).
Writing out leftSubcatModel...done (00:02.897).
Writing out rightSubcatModel...done (00:01.551).
Writing out modNonterminalModel...done (00:13.506).
Writing out modWordModel...done (00:12.669).
Writing out rightSubcatMap...done (00:00.040).
Writing out modNonterminalMap...done (00:00.173).
Writing out simpleModNonterminalMap...done (00:00.047).
Writing out prunedPreterms...done (00:00.000).
Writing out prunedPunctuation...done (00:00.000).
Finished outputting derived counts in 00:37.768.

Total elapsed time: 02:48.715.

Have a nice day!

I feel like writing to Dan: "Thanks! you too!"