Multilingual Sentiment Analysis

def prune_transformer(model, amount=0.3): for name, module in model.named_modules(): if isinstance(module, torch.nn.Linear): prune.l1_unstructured(module, name='weight', amount=amount) prune.remove(module, 'weight') return model # Efficient Multilingual NMT class LightTranslator(nn.Module): def __init__(self, vocab_size, d_model=512): super().__init__() self.encoder = TransformerEncoder(d_model) self.decoder = TransformerDecoder(d_model) def forward(self, src, tgt): memory = self.encoder(src) return self.decoder(tgt, memory)

Project Overview

Extended sentiment classification work beyond Arabic to a multilingual setting, evaluating how well transformer-based approaches generalize across languages with different morphological structures and resource levels.

Technical Details

Architectures: Multilingual transformer models (e.g. mBERT, XLM-R) evaluated against language-specific fine-tuned models.
Languages: Arabic, French, and English — deliberately mixing related, well-resourced pairs with a morphologically distant, lower-resource one.
Approach: Compared zero-shot cross-lingual transfer against per-language fine-tuning, to assess how much language-specific adaptation is actually needed.