Giter Site home page Giter Site logo

khmernlp's Introduction

DOI

KCC-Level Network

Khmer Word Segmentation

Word segmentation on Khmer texts is a challenging task since in Khmer texts, there is no explicit word delimiters such as a space. This is not the case for Latin languages such French or English. In Khmer texts, characters are written from left to right consecutively with optional space between words. Another challenge is that most words can be co-located to form a new word. For example, សម្តេច (your highness) can be a proper word by itself or can be split into two words: ស(white)and ម្តេច (how) both of which are proper words by themselves [1; 2; 3].

Khmer Character Cluster(KCC)

The concept of Khmer Character Cluster (KCC) was introduced in [3; 6]. KCC is the inseparable sequence of characters. In Khmer writing system, a vowel cannot be by itself; a vowel must be placed after a consonant. Here are a few examples showing that a word is a combination of KCCs [6]:

  • សាលាក្តី has 3 KCCs (សា+លា+ក្តី)
  • ចុកចាប់ has 4 KCCs (ចុ+ក+ចា+ប់)
  • ស្ត្រី has 1 KCC only.

KCC Rules

BiLSTM Networks for Khmer Word Segmentation

Character Level vs KCC Level

Character Level

Character-Level Network

KCC Level

KCC-Level Network

Word Segmentation of Khmer Text Using Conditional Random Fields

For CRF model for work segmentation, reader is referred to the works by [1] and [5]. The pre-trained CRF model used in the package is obtained from Phylypo Tum (https://medium.com/@phylypo/segmentation-of-khmer-text-using-conditional-random-fields-3a2d4d73956a). Phylypo Tum extends the work by [1] by introducing KCC instead of character level modelling.

Running Word Segmentation Usign Pre-Trained KCC Network

The easy way to run the segmentation models without worrying about package dependencies is to open 'Run_segmentation_colab.ipynb' in Google Colab.

from khmerwordsegmentor import KhmerWordSegmentor
seg = KhmerWordSegmentor()
ts = "ចំណែកជើងទី២ នឹងត្រូវធ្វើឡើងឯប្រទេសកាតា៕"
print('Segmention by LSTM: ', seg.segment(ts,model='lstm'))
print('Segmention by CRF: ', seg.segment(ts,model='crf'))


#Inference on GPU!
#Segmention by LSTM:  ចំណែក-ជើង-ទី-២-នឹង-ត្រូវ-ធ្វើឡើង-ឯ-ប្រទេស-កាតា-៕
#Segmention by CRF:  ចំណែកជើង-ទី-២-នឹង-ត្រូវ-ធ្វើឡើង-ឯ-ប្រទេស-កាតា-៕

TODO

  • Character Level Model
  • Model Training
  • Khmer Word Embedding

Citation

@misc{RinaB2020,
  author = {Rina Buoy and Sokchea Kor},
  title = {Khmer Word Segmentation Using BiLSTM Networks},
  year = {2020},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/rinabuoy/KhmerNLP}}
}

References

[1] Vichet Chea, Ye Kyaw Thu, Chenchen Ding, Masao Utiyama, Andrew Finch, and Eiichiro Sumita. Khmer word segmentation using conditional random fields. Khmer Natural Language Processing, 2015.

[2] Narin Bi and Nguonly Taing. Khmer word segmentation based on bidirectional maximal matching for plaintextand microsoft word. Signal and Information Processing Association Annual Summit and Conference (APSIPA), 2014.

[3] Chea Sok Huor, Top Rithy, Ros Pich Hemy, Vann Navy, Chin Chanthirith, and Chhoeun Tola. Word bigram vs orthographic syllable bigram in khmer word. PAN Localization Team, 2007.

[4] Ye Kyaw Thu, Vichet Chea, Andrew Finch, Masao Utiyama, and Eiichiro Sumita. A large-scale study of statistical machine translation methods for Khmer language. In Proceedings of the 29th Pacific Asia Conference on Language, Information and Computation, pages 259–269, Shanghai, China, October 2015.

[5] Phylypo Tum. Word Segmentation of Khmer Text Using Conditional Random Fields, June 2020.

[6] Chea Sok Huor and Top Rithy. Detection and correction of homophonous error word for khmer language. 2007.

[7] Andrej Karpathy, Justin Johnson, and Li Fei-Fei. Visualizing and understanding recurrent networks, 2015.

[8] Zhiheng Huang, Wei Xu, and Kai Yu. Bidirectional lstm-crf models for sequence tagging, 2015.

khmernlp's People

Contributors

rinabuoy avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.