Transferred Embeddings for Igbo Similarity, Analogy, and Diacritic Restoration Tasks

Ignatius Ezeani, Ikechukwu Onyenwe, Mark Hepple


Abstract
Existing NLP models are mostly trained with data from well-resourced languages. Most minority languages face the challenge of lack of resources - data and technologies - for NLP research. Building these resources from scratch for each minority language will be very expensive, time-consuming and amount largely to unnecessarily re-inventing the wheel. In this paper, we applied transfer learning techniques to create Igbo word embeddings from a variety of existing English trained embeddings. Transfer learning methods were also used to build standard datasets for Igbo word similarity and analogy tasks for intrinsic evaluation of embeddings. These projected embeddings were also applied to diacritic restoration task. Our results indicate that the projected models not only outperform the trained ones on the semantic-based tasks of analogy, word-similarity, and odd-word identifying, but they also achieve enhanced performance on the diacritic restoration with learned diacritic embeddings.
Anthology ID:
W18-4004
Volume:
Proceedings of the Third Workshop on Semantic Deep Learning
Month:
August
Year:
2018
Address:
Santa Fe, New Mexico
Editors:
Luis Espinosa Anke, Dagmar Gromann, Thierry Declerck
Venue:
SemDeep
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
30–38
Language:
URL:
https://aclanthology.org/W18-4004
DOI:
Bibkey:
Cite (ACL):
Ignatius Ezeani, Ikechukwu Onyenwe, and Mark Hepple. 2018. Transferred Embeddings for Igbo Similarity, Analogy, and Diacritic Restoration Tasks. In Proceedings of the Third Workshop on Semantic Deep Learning, pages 30–38, Santa Fe, New Mexico. Association for Computational Linguistics.
Cite (Informal):
Transferred Embeddings for Igbo Similarity, Analogy, and Diacritic Restoration Tasks (Ezeani et al., SemDeep 2018)
Copy Citation:
PDF:
https://aclanthology.org/W18-4004.pdf