PUBLICATION: “Something Something Hota Hai ”, - An Explainable Approach towards Sentiment Analysis on Indian Code-Mixed Data
accepted at Workshop on Noisy User-generated Text (W-NUT), EMNLP 2021.
discusses the implications model-agnostic explainable methods(Lime and Shap) for hindi-english code-mixed data. Models compared - SVM, Random Forest, CNN, CNN with Glove Embeddings, LSTM, LSTM with Glove Embeddings, XLMRoBERTa.