Contextual Text Embeddings for Twi
arXiv:2103.15963
Abstract
Transformer-based language models have been changing the modern Natural Language Processing (NLP) landscape for high-resource languages such as English, Chinese, Russian, etc. However, this technology does not yet exist for any Ghanaian language. In this paper, we introduce the first of such models for Twi or Akan, the most widely spoken Ghanaian language. The specific contribution of this research work is the development of several pretrained transformer language models for the Akuapem and Asante dialects of Twi, paving the way for advances in application areas such as Named Entity Recognition (NER), Neural Machine Translation (NMT), Sentiment Analysis (SA) and Part-of-Speech (POS) tagging. Specifically, we introduce four different flavours of ABENA -- A BERT model Now in Akan that is fine-tuned on a set of Akan corpora, and BAKO - BERT with Akan Knowledge only, which is trained from scratch. We open-source the model through the Hugging Face model hub and demonstrate its use via a simple sentiment classification example.
10 pages paper; Accepted at African NLP Workshop @ EACL 2021
References in corpus (12)
- Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling
- word2vec Explained: deriving Mikolov et al.'s negative-sampling word-embedding method
- An Introduction to Convolutional Neural Networks
- Exploring the Limits of Language Modeling
- Unsupervised Learning of Sentence Embeddings using Compositional n-Gram Features
- An Analysis of Neural Language Modeling at Multiple Scales
- Transfer Learning across Low-Resource, Related Languages for Neural Machine Translation
- A Neural Knowledge Language Model
- Unsupervised Cross-Modal Alignment of Speech and Text Embedding Spaces
- An Ensemble Method to Produce High-Quality Word Embeddings (2016)
- Masakhane -- Machine Translation For Africa
- Word2Vec and Doc2Vec in Unsupervised Sentiment Analysis of Clinical Discharge Summaries