Twitter is an important source of information that people frequently contribute to and rely on for emerging topics, public opinions, and event awareness. Crisis-relevant tweets can potentially avail a magnitude of applications such as helping authorities and governments become aware of situations and thus offer better responses. One major challenge toward crisis-awareness in Twitter is to identify those tweets that are relevant to unseen crises. In this article, we propose an automatic labeling approach to distinguishing crisis-relevant tweets while differentiating source types (e.g., government or personal accounts) simultaneously. We first analyze and identify tweet-specific linguistic, sentimental, and emotional features based on statistical topic modeling. Then, we design a novel correlative convolutional neural network which uses a shared hidden layer to learn effective representations of the multi-faceted features. The model can discover salient information while being robust to the variations and noises in tweets and sources. To obtain a bird’s-eye view of a crisis event, we further develop an approach to automatically summarize key information of identified tweets. Empirical evaluation on a real Twitter dataset demonstrates the feasibility of discerning relevant tweets for an unseen crisis. The applicability of our proposed approach is further demonstrated with a crisis aider system.

The global urbanization imposes unprecedented pressure on urban infrastructure and public resources. The population explosion has made it challenging to satisfy the daily needs of urban residents. 'Smart City' is a solution that utilizes different types of data collection sensors to help manage assets and resources intelligently and more efficiently. Under the Smart City umbrella, the primary research initiative in improving the efficiency of car-hailing services is to predict the citywide passenger demand to address the imbalance between the demand and supply. However, predicting the passenger demand requires analysis on various data such as historical passenger demand, crowd outflow, and weather information, and it remains challenging to discover the latent relationships among these data. To address this challenge, we propose to improve the passenger demand prediction via learning the salient spatial-temporal dynamics within a reinforcement learning framework. Our model employs an information selection mechanism to focus on the most distinctive data in historical observations. This mechanism can automatically adjust the information zone according to the prediction performance to find the optimal choice. It also ensures the prediction model to take full advantage of the available data by introducing the positive and excluding the negative correlations. We have conducted experiments on a large-scale real-world dataset that covers 1.5 million people in a major city in China. The results show our model outperforms state-of-the-art and a series of baselines by a large margin.