Attention Based LSTM CNN Framework for Sentiment Extraction from Bengali Texts

Sentiment analysis is a field of immense possibilities and application despite being an age old topic. Variants of neural networks with attention mechanism is a well known tool in this field. As many people now use Bengali to express their opinion in online platforms, it becomes inevitable to analyze emotions in social networks and product reviews in online commerce sites. But unfortunately very few of these techniques have been applied for Bengali sentences. In this paper an attention based LSTM-CNN model is being proposed to be used to solve the problem which outperforms all the latest neural network based approaches in terms of accuracy and model complexity.