NOTE: tensorflow/syntaxnet now includes Parsey Universal that provides models for other languages, including Portuguese, available here.

syntaxnet-portuguese

A Portuguese language model for syntaxnet, treebank data used is available from Universal Dependencies.

Usage

Clone and build syntaxnet, then:

$ git clone git@github.com:nunorc/syntaxnet-portuguese.git
$ cd syntaxnet-portuguese
# edit test.sh to update your syntaxnet home directoy
$ echo "A Maria tem razão."  | ./test.sh 
(...)
INFO:tensorflow:Read 1 documents
Input: A Maria tem razão .
Parse:
tem VERB ROOT
 +-- Maria PROPN nsubj
 |   +-- A DET det
 +-- razão NOUN dobj
 +-- . PUNCT punct

Acknowledgements

Thank you to tensorflow and syntaxnet developers and maintainers.

Thank you to Myungchul Shin for sharing his test code.