https://github.com/nibircse/BengaliDependencyParser

https://github.com/nibircse/BengaliDependencyParser

Last synced: about 1 month ago
JSON representation

• Host: GitHub
• URL: https://github.com/nibircse/BengaliDependencyParser
• Owner: nibircse
• Created: 2016-08-23T14:55:41.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2014-05-03T04:04:04.000Z (about 10 years ago)
• Last Synced: 2024-05-20T20:44:50.077Z (about 1 month ago)
• Language: C++
• Size: 13.4 MB
• Stars: 4
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Lists

• awesome-bangla - Bengali Dependency Parser

README

        
BengaliDependencyParser

=======================

This is implementation of a dependency parser of the language Bengali. This repository contains a dataset of around 7500 annotated tokens of Bengali Text. For the annotation decisions (and caveats) please refer the report. 

Contents

Data/ - Contains the annotated dataset. All the files are in CoNLL format.

Data/Train - The training set (total of 5463 tokens annotated)

Data/Train/train.txt.conll - the annotated train file

Data/Test/TestA/testA.conll - the annotated test file

Data/Test/TestA/testB.conll - the annotated test file

To run the parser:

You can use your own training data or use the one shared my me. To run do the following.

 cd TurboParser-2.1.0.

Train script: ./run_train.sh 

Test script : ./run_test.sh 

It will run all the 3 (basic, standard and full models) of Turbo Parser. The labelled/unlabelled accuracy will be printed on the console after the test script is run.

Thanks!