URBAN: Universal Rule-based Machine Translation NLP Toolkit
Project description
URBAN: Universal Rule-Based Machine Translation toolkit
A tool for translating text from source grammar to target grammar (context-free) with corresponding dictionary.
Why not translate it yourself when Google Translate cannot satisfy you❓
⚙️ Installation
pip install urban
✨ What is good about URBAN?
- Rule-based, deterministic translation; unlike Google Translate - giving only 1 non-deterministic result
- Using NLTK parsing interface and is built on top of already-efficient NLTK backend
- Can be used for data augmentation
📖 Usage
from urban import Translator
# Source sentence to be translated
src_sentences = ["I love good dogs", "I hate bad dogs"]
# Source grammar in nltk parsing style
src_grammar = """
S -> NP VP
NP -> PRP
VP -> VB NP
NP -> JJ NN
PRP -> 'I'
VB -> 'love' | 'hate'
JJ -> 'good' | 'bad'
NN -> 'dogs'
"""
# Some edit within source grammar to target grammar
src_to_target_grammar = {
"NP -> JJ NN": "NP -> NN JJ" # in Vietnamese NN goes before JJ
}
# Word-by-word dictionary from source language to target language
en_to_vi_dict = {
"I":"tôi",
"love":"yêu",
"hate":"ghét",
"dogs":"những chú_chó",
"good":"ngoan",
"bad":"hư"
}
translator = Translator(src_grammar = src_grammar,
src_to_tgt_grammar = src_to_target_grammar,
src_to_tgt_dictionary = en_to_vi_dict)
trans_sentences = translator.translate(src_sentences)
# This should returns ['tôi yêu những chú_chó ngoan', 'tôi ghét những chú_chó hư']
⚖️ License
This repository is using the Apache 2.0 license that is listed in the repo. Please take a look at LICENSE
as you wish.
✍️ BibTeX
If you wish to cite the framework feel free to use this (but only if you loved it 😊):
@misc{phat2020urban,
author = {Truong-Phat Nguyen},
title = {URBAN: Universal Rule-Based Machine Translation NLP toolkit},
year = {2020},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {\url{https://github.com/nlp-urban/urban}},
}
Contributors:
- Patrick Phat Nguyen
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
ruby-0.0.1.tar.gz
(5.5 kB
view hashes)
Built Distribution
ruby-0.0.1-py3-none-any.whl
(10.3 kB
view hashes)