Attention mechanism-based news sentiment analyzer

Sentiment Analysis is the task of determining the feeling or opinion of a chunk of text. One of the crucial applications of sentiment analysis is to classify the various news articles into three fundamental categories of sentiments—negative, positive, and neutral. By considering the sentiments of news articles, we can figure out whether the writer’s sentiment is negatively or positively oriented. Multiple models have been constructed to analyze the news articles but none of them compassed at both the levels of a sentence as well as the whole document. In this paper, I have proposed a method which can perform this task at both the levels with accurate results, by using LSTM network and a deep learning framework, i.e., attention network.