Abstract:
Sentiment analysis is one of the most widely studied applications in NLP, but most work focuses on languages with large amounts of data. We introduce the first large-scale human-annotated Twitter sentiment dataset for the four most widely spoken languages in Nigeria—Hausa, Igbo, Nigerian-Pidgin, and Yorùbá—consisting of around 30,000 annotated tweets per language (and 14,000 for Nigerian-Pidgin), including a significant fraction of code-mixed tweets. We propose text collection, filtering, processing, and labelling methods that enable us to create datasets for these low-resource languages. We evaluate a range of pre-trained models and transfer strategies on the dataset. We find that language-specific models and language-adaptive fine-tuning generally perform best. We release the datasets, trained models, sentiment lexicons, and code to incentivize research on sentiment analysis in under-represented languages.
Bio:
Shamsuddeen is a PhD candidate at the MAPi-Joint Doctoral Program in Computer Science at the University of Porto. His current research interests focus on natural language processing for low-resource languages. He received his Master's degree from the University of Manchester, UK, and a Bachelor's Degree from Bayero University, Kano, Nigeria. He is a member of MasakhaneNLP and a faculty member at the Faculty of Computing at Bayero University, Kano, Nigeria.
Join the talk here on MS Teams
The talk is on creating Nigerian sentiment corpus. Paper submitted to LREC and under review: https://arxiv.org/pdf/2201.08277.pdf