Sentiment and Semantic Embeddings for Disaster Decision-Making

The popularity of social media and public opinions in online news channels are attracting the attention of government to help for finding out better private/public companies for efficient disaster management. The obtained accuracy was around 60% due to the lack of information about the company performance during disaster management. Relying on polarity estimation, our aim is to predict the company’s performance as good or bad. Now, we have introduced a centralized tool for classifying the companies performance during disasters using the KNN, Doc2vec and Deep CNN approach. The accuracy of this model is approx 93%, for online datasets.