Agha Ali Raza

Publications/Natural Language Processing & LLMs for Low-Resource Languages

Conference · 2020

SimplifyUR: Unsupervised Lexical Text Simplification for Urdu

Namoos Hayat Qasmi, Haris Bin Zia, Awais Athar, Agha Ali Raza

LREC 2020, Marseille, France

Summary

Simplifying text makes it accessible to more readers, but existing methods depend on resources that Urdu, spoken by 170 million people, does not have. This paper presents the first automatic text simplification system for Urdu. It is unsupervised: it needs only plain Urdu text, and uses word embeddings and morphological features to replace complex words with simpler ones while handling inflection and honorifics properly. The system scores 80.15 BLEU and 42.02 SARI against manually simplified text, and human judges assess its output for correctness, grammaticality, meaning preservation, and simplicity. Code and corpus are public.

Cite this paper

Namoos Hayat Qasmi, Haris Bin Zia, Awais Athar, Agha Ali Raza. SimplifyUR: Unsupervised Lexical Text Simplification for Urdu. LREC 2020, Marseille, France.

@inproceedings{qasmi2020simplifyur,
  author = {Namoos Hayat Qasmi and Haris Bin Zia and Awais Athar and Agha Ali Raza},
  title = {{SimplifyUR: Unsupervised Lexical Text Simplification for Urdu}},
  booktitle = {LREC 2020, Marseille, France},
  year = {2020},
  url = {https://aghaaliraza.com/publications/2020--LREC--SimplifyUR-Unsupervised_Lexical_Text_Simplification_for_Urdu.pdf}
}

More on this theme

All 18 works on this theme →