Research & Publications

Selected papers, reproducible code, and shared datasets.

Advancing Language Diversity and Inclusion: Towards a Neural Network-based Spell Checker and Correction for Wolof

Thierno Ibrahima Cissé, Fatiha Sadat • Proceedings of the Fifth Workshop on Resources for African Indigenous Languages @ LREC-COLING • 2024

Introduces a neural spell-checking approach for Wolof using transformer models trained on semi-automatically generated pairs of misspelled and corrected sentences. The paper reports promising results for improving text quality in a low-resource language setting and contributes to more accessible digital communication for under-represented languages.

PaperCode

Automatic Spell Checker and Correction for Under-represented Spoken Languages: Case Study on Wolof

Thierno Ibrahima Cissé, Fatiha Sadat • Proceedings of the Fourth Workshop on Resources for African Indigenous Languages @ ACL • 2023

Presents a Wolof spell checker built with a trie, dynamic programming, and weighted Levenshtein distance, supported by newly created lexical resources and a corpus of misspellings. Despite limited training data, the system achieved 98.31% predictive accuracy and 93.33% suggestion accuracy, providing a strong foundation for future Wolof language tools.

PaperCode