Classification of Precious and Non-precious Tweets Using Deep Learning

Social media is used to pass information and send messages during many situations. Twitter can be used as primary source for sharing real-time information during disasters. In disaster situations, users may send many tweets on Twitter. Detecting precious and non-precious tweets is important during disaster situations. To identify precious tweets, image features and text features are meaningful. The current methods uses text features and image features to classify precious and non-precious tweets. The traditional approach’s limitations are heavier in the model, take time for training, and vanishes the gradient problem. We propose different models for image tweets and textual tweets in the classification of precious and non-precious tweets. Immense experiments are carried out on Twitter dataset during various disasters.