Question-answering models for Arabic, Bengali and Indonesian languages.
- 1 minSee the GitHub repository here.
You can read the paper here.
For the Natural Language Processing class at the University of Copenhagen, two students and I collaborated on building different question-answering models for Arabic, Bengali and Indonesian texts, and testing the results.
We experienced with:
- Rule-based classifier
- Bigram Language Model
- LSTM
- RNN
- Logistic regression model with BPEmb and TF-ID features
- Different Transformer models