Urdu Sentiment Analysis Using Deep Attention-Based Technique

Abstract

Sentiment analysis (SA) is a process that aims to classify text into positive, negative or neutral categories. It has recently gained the research community's attentionbecause of the abundance of opinion data on the internet. Deep learning techniques are widely used for language processing but are seen as black boxes, and their effectiveness comes ininterpretability. The major goal of this article is to create an Urdu SA model that can comprehend review semantics without the need of language resources. Wedesign an attention-based neural network for the review level Urdu SA. For better results, we used atransfer learning approach that uses pre-trained embedding’s. The Visualization of attention weights isalso measuredthat uncovers the black box of the models and confirms their intuition, which aids in the interpretation of the model's learned representations. The proposed model is tested and evaluated in terms of accuracy and F1 score. The proposed model archives 91% accuracy and 88% F1 score,respectively.