Read the issue description on github. I added possible false alarms there.
Regards
Marcin
07-10-2013 11:11, "Kumara Bhikkhu" <[email protected]> napisał(a):
> Thanks for trying to explain, Marcin. Like I said, I don't have a good
> head for this stuff. Examples work better for me. So, I think I'll get a
> better idea when I see the false alarms. First, I need to make the rule
> match the incorrect sentence.
>
> Perhaps "a dictionary of phrasal verbs" wouldn't work either, as the
> possibilities are endless. Consider these correct sentences: I *want to
> break* free. (Freddie Mercury) You *are to report* at 5am. It *is*important
> *to make* this explicit. I've *been trying to call* you. She *wants* you *to
> go*. He *has been wanting *very much *to meet* you. She *is manipulating*her
> father
> *to get *her way. He *tried* very hard *to lift* the rock. I'm *going* to
> the hardware store *to buy* a hammer. (Not sure if this one would be a
> problem.) Perhaps we can ignore the concept of phrasal verbs, and just
> match the last word of the first phrasal verb before "to", besides the verb
> right after "to".
>
> Anyway, I just realised that I missed out postag_regexp="yes" for the
> last token. :-p
>
> After some testing, I realised that to avoid false alarms, I can only have
> the rule flag "to VBG|VBZ". Here's the result:
>
> <rule id="TO_VBG|VBZ" name="to gerund or 3rd person singular">
> <pattern>
> ** **** **<token postag_regexp="yes"
> postag="VBD|VBG|VBN|VBP|VBZ"/>
> ** **** **<token min="0"/>
> ** **** ** <token>to</token>
> ** **** **<marker>
> ** **** **** **<token postag_regexp="yes"
> postag="VBG|VBZ"></token>
> ** **** ** </marker>
> ** ** </pattern>
> ** ** <message>You might need to use the base form of the verb
> here: <suggestion><match no="4" postag="VB"></match></suggestion>.</message>
> ** ** <short>To + infinitive</short>
> ** ** <example type="correct">I was surprised to
> <marker>learn</marker> this.</example>
> ** ** <example type="correct">He spoke to chosen people.</example>
> ** ** <example type="correct">You are to report at 5am.</example>
> ** ** <example type="correct">It is important to make this
> explicit.</example>
> ** ** <example type="correct">I've been trying to call
> you.</example>
> ** ** <example type="correct">She wants you to go. </example>
> ** ** <example type="correct">He has been wanting very much to
> meet you.</example>
> ** ** <example correction="learn" type="incorrect">I was surprised
> to <marker>learns</marker> this.</example>
> ** ** <example correction="get" type="incorrect">he is
> manipulating her father to <marker>getting</marker> her way.</example>
> </rule>
>
> Marcin Miłkowski wrote thus at 06:18 PM 06-10-13:
>
> Actually, you're trying to achieve something very difficult -- the
> trouble is that we do no deep parsing, which would assign verb groups
> correctly.
>
> I created an issue for this to remind us about it:
>
> https://github.com/languagetool-org/languagetool/issues/20
>
> But unless we have a dictionary of phrasal verbs that require 'to' and
> get no infinitive, we will surely create a lot of false alarms.
>
> Best,
> Marcin
>
>
>
> ------------------------------------------------------------------------------
> October Webinars: Code for Performance
> Free Intel webinars can help you accelerate application performance.
> Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most
> from
> the latest Intel processors and coprocessors. See abstracts and register >
> http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk
> _______________________________________________
> Languagetool-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/languagetool-devel
>
>
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk
_______________________________________________
Languagetool-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/languagetool-devel
