A Neural Network Model to Identify the Crisis-related Actionable Informative Tweets for Disaster Management

During emergencies and disaster situations, we can consider twitter as a source of situation related information. Identifying informative tweets from twitter streams provide enormous opportunities for public safety personnel in coordinating aid operations or post-incident assessment. However, the brevity of tweets and its' noisy contents make it challenging to extract the necessary information effectively and identify the tweets based on different information types. In this paper, we present a unified neural architecture for actionable informative tweet classification. In our proposed model, we exploit the transfer learning features from a pre-trained sentence embedding model along with a rich set of hand-crafted features to train a multilayer perceptron (MLP) network. Besides, we employ the state-of-the-art LSTM variants, nested LSTMs (NLSTMs) to capture the long-term dependency effectively. On top of nested LSTMs, we perform the convolution using multiple kernels (CMK) to obtain the higher-level representation of tweets. Experiments on 2018 TREC incident streams (TREC-IS) dataset show that our proposed neural model gains a significant improvement against the current state-of-the-art.