The language model ParsPer (Seraji, 2015) is developed using the graph-based MateParser (Bohnet, 2010), an open source statistical dependency parser in the NLP toolkit. ParsPer is created by training the graph-based MateParser on the entire Uppsala Persian Dependency Treebank (UPDT) with a selected configuration. For detailed description, see Chapter 6, pp. 147-172 in Seraji (2015).
The tool is developed by Mojgan Seraji ( mojgan.seraji96@gmail.com ) and licensed under GNU General Public License. The tools is used for parsing Persian texts and can be downloaded below:
Before you start using the language model, you will first need to download anna-3.61.jar. Then you can run the ParsPer by typing the following at the command line prompt:
prompt> java -cp anna-3.61.jar is2.parser.Parser -model model_ParsPer -test input.conll -out output.conll
1. Bohnet, Bernd (2010). Top Accuracy and Fast Dependency Parsing is not a Contradiction. In Proceedings of the 23rd International Conference on Computational Linguistics (Coling 2010), pp. 89–97.
2. Seraji, Mojgan. 2015. Morphosyntactic Corpora and Tools for Persian. Doctoral dissertation, Uppsala University. Studia Linguistica Upsaliensia 16. [pdf]