Sentiment-Analysis-for-Code-Mixed-Data

This is the repository for the paper 'Transfer Learning for Code-Mixed Data: Do Pretraining Languages Matter?' It contains links to all the datasets used in the paper.

Original Datasets

• AfriSenti - train, dev
• NaijaVader - in this repo
• SAIL
• IIITH-CodeMix
• TamilMixSentiment
• MalayalamMixSentiment
• DravidianCodeMix
• Singh_et_al (NER)
• MasakhaNER
• Shah_and_Maurya (Sarcasm Detection)

Modified versions of TamilMixSentiment, MalayalamMixSentiment and DravidianMixSentiment (transliterated and with 4 sentiment labels) are included in this repo. Splits that we made for IIITH-CodeMix and Singh_et_al (NER) are also included.

Code

• To reproduce the experiments in the paper, please refer to the MaChAmp repo.