KaustubhSharedTask@LT-EDI 2023: Homophobia-Transphobia Detection in Social Media Comments with NLPAUG-driven Data Augmentation

Kaustubh Lande, Rahul Ponnusamy, Prasanna Kumar Kumaresan, Bharathi Raja Chakravarthi


Abstract
Our research in Natural Language Processing (NLP) aims to detect hate speech comments specifically targeted at the LGBTQ+ community within the YouTube platform shared task conducted by LTEDI workshop. The dataset provided by the organizers exhibited a high degree of class imbalance, and to mitigate this, we employed NLPAUG, a data augmentation library. We employed several classification methods and reported the results using recall, precision, and F1-score metrics. The classification models discussed in this paper include a Bidirectional Long Short-Term Memory (BiLSTM) model trained with Word2Vec embeddings, a BiLSTM model trained with Twitter GloVe embeddings, transformer models such as BERT, DistiBERT, RoBERTa, and XLM-RoBERTa, all of which were trained and fine-tuned. We achieved a weighted F1-score of 0.699 on the test data and secured fifth place in task B with 7 classes for the English language.
Anthology ID:
2023.ltedi-1.10
Volume:
Proceedings of the Third Workshop on Language Technology for Equality, Diversity and Inclusion
Month:
September
Year:
2023
Address:
Varna, Bulgaria
Editors:
Bharathi R. Chakravarthi, B. Bharathi, Joephine Griffith, Kalika Bali, Paul Buitelaar
Venues:
LTEDI | WS
SIG:
Publisher:
INCOMA Ltd., Shoumen, Bulgaria
Note:
Pages:
71–77
Language:
URL:
https://aclanthology.org/2023.ltedi-1.10
DOI:
Bibkey:
Cite (ACL):
Kaustubh Lande, Rahul Ponnusamy, Prasanna Kumar Kumaresan, and Bharathi Raja Chakravarthi. 2023. KaustubhSharedTask@LT-EDI 2023: Homophobia-Transphobia Detection in Social Media Comments with NLPAUG-driven Data Augmentation. In Proceedings of the Third Workshop on Language Technology for Equality, Diversity and Inclusion, pages 71–77, Varna, Bulgaria. INCOMA Ltd., Shoumen, Bulgaria.
Cite (Informal):
KaustubhSharedTask@LT-EDI 2023: Homophobia-Transphobia Detection in Social Media Comments with NLPAUG-driven Data Augmentation (Lande et al., LTEDI-WS 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.ltedi-1.10.pdf