SRL not works with some sentences

[cesarcruz]

I installed nlpnet and downloaded data. But it works only with some sentences. I use Ubuntu 13.04 32bits and Python 2.7.4 with virtualenv.

Here I posted two examples, the first seems to work, the second no:

Example 1 (Seem to work):

python nlpnet-tag.py srl /path/to/nlpnet-data/
This is the first sentence of the paragraph. This is the second sentence of the paragraph.
This is the first sentence of the paragraph .
sentence
    A1: the paragraph
    A0: first
    V: sentence

This is the second sentence of the paragraph .
sentence
    A1: the paragraph
    A0: second
    V: sentence

Example 2 (Not works):

This sentence is taked from CoNLL-2005 Examlples

python nlpnet-tag.py srl /path/to/nlpnet-data/
The $1.4 billion robot spacecraft faces a six-year journey to explore Jupiter and its 16 known moons.
The $ 1.4 billion robot spacecraft faces a six-year journey to explore Jupiter and its 16 known moons .

I also tested from python and tokenization works but in the second example calling arg_structures, returns an empty list.

Please let me know if you need something more to address this issue.

Thanks

[erickrf]

Hi, the problem here is that the trained models from http://nilc.icmc.usp.br/nilc/download/nlpnet-data.zip were trained for Portuguese. It won't work in English.

When you get an empty arg_structures list, it means that the network didn't recognize any token as a predicate.

[cesarcruz]

@erickrf thanks for your answer, it is clear for me now, but could you let me know where to find trained models for English that work with nlpnet? ...I like this package and want to use it in my project but for english language, thanks.

[cesarcruz]

Oh, sorry, I see issue #1 now which seems explain that.