@inproceedings{shanmugavadivel-etal-2025-deathly-hallows,
    title = "{T}he{\_}{D}eathly{\_}{H}allows@{D}ravidian{L}ang{T}ech 2025: Multimodal Hate Speech Detection in {D}ravidian Languages",
    author = "Shanmugavadivel, Kogilavani  and
      Subramanian, Malliga  and
      K, Vasantharan  and
      A, Prethish G  and
      S, Santhosh",
    editor = "Chakravarthi, Bharathi Raja  and
      Priyadharshini, Ruba  and
      Madasamy, Anand Kumar  and
      Thavareesan, Sajeetha  and
      Sherly, Elizabeth  and
      Rajiakodi, Saranya  and
      Palani, Balasubramanian  and
      Subramanian, Malliga  and
      Cn, Subalalitha  and
      Chinnappa, Dhivya",
    booktitle = "Proceedings of the Fifth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages",
    month = may,
    year = "2025",
    address = "Acoma, The Albuquerque Convention Center, Albuquerque, New Mexico",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/moar-dois/2025.dravidianlangtech-1.113/",
    doi = "10.18653/v1/2025.dravidianlangtech-1.113",
    pages = "661--665",
    ISBN = "979-8-89176-228-2",
    abstract = "The DravidianLangTech@NAACL 2025 shared task focused on multimodal hate speech detection in Tamil, Telugu, and Malayalam using social media text and audio. Our approach integrated advanced preprocessing, feature extraction, and deep learning models. For text, preprocessing steps included normalization, tokenization, stopword removal, and data augmentation. Feature extraction was performed using TF-IDF, Count Vectorizer, BERT-base-multilingual-cased, XLM-Roberta-Base, and XLM-Roberta-Large, with the latter achieving the best performance. The models attained training accuracies of 83{\%} (Tamil), 88{\%} (Telugu), and 85{\%} (Malayalam). For audio, Mel Frequency Cepstral Coefficients (MFCCs) were extracted and enhanced with augmentation techniques such as noise addition, time-stretching, and pitch-shifting. A CNN-based model achieved training accuracies of 88{\%} (Tamil), 88{\%} (Telugu), and 93{\%} (Malayalam). Macro F1 scores ranked Tamil 3rd (0.6438), Telugu 15th (0.1559), and Malayalam 12th (0.3016). Our study highlights the effectiveness of text-audio fusion in hate speech detection and underscores the importance of preprocessing, multimodal techniques, and feature augmentation in addressing hate speech on social media."
}