W dniu 2014-09-03 10:57, Jaume Ortolà i Font pisze:
> 2014-09-03 9:59 GMT+02:00 Marcin Miłkowski <list-addr...@wp.pl
> <mailto:list-addr...@wp.pl>>:
>
>
> We could, in principle, try to add this kind of test to the
> disambiguator action but I'm not sure if it won't break something.
>
>
> Hi,
>
> I agree with Dominique. This behavior can generate errors which are very
> hard to detect.
>
> If we change it, it's more than probable than something will break, as
> we have a lot of disambiguator rules now. Perhaps some filter action is
> to be replaced by a replace action. But I think it's worth a try.
As I expected, it's not trivial. It works fine for Polish and English
but not for Catalan. There must be a filter somewhere in the Catalan
disambiguator file that expects this behavior. To test, please use this
code in DisambiguationPatternRuleReplacer.java, from line 346:
boolean newPOSmatches = false;
// only apply filter rule when it matches previous tags:
for (int i = 0; i < whTokens[fromPos].getReadingsLength(); i++) {
if (!whTokens[fromPos].getAnalyzedToken(i).hasNoTag() &&
whTokens[fromPos].getAnalyzedToken(i).getPOSTag().matches(disambiguatedPOS))
{
newPOSmatches = true;
break;
}
}
if (newPOSmatches) {
final MatchState matchState =
tmpMatchToken.createState(rule.getLanguage().getSynthesizer(),
whTokens[fromPos]);
final String prevValue = whTokens[fromPos].toString();
final String prevAnot =
whTokens[fromPos].getHistoricalAnnotations();
whTokens[fromPos] = matchState.filterReadings();
annotateChange(whTokens[fromPos], prevValue, prevAnot);
}
---
(replace everything until the break statement).
You will see that the Catalan pattern rule breaks then. Please fix it,
and I'll see if that's everything we need.
Regards,
Marcin
------------------------------------------------------------------------------
Slashdot TV.
Video for Nerds. Stuff that matters.
http://tv.slashdot.org/
_______________________________________________
Languagetool-devel mailing list
Languagetool-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/languagetool-devel
