Automatic Labeling Streaming Data for Event Detection from Heterogeneous Sensors

In this paper, we introduce a new method for detecting events by automatically labeling streaming data from heterogeneous sensors. The proposed method argues that by finding spatio-temporal-theme correlations between physical and social sensory data, events can be detected precisely and explained understandably in real-time. Here, a training stage is designed as a non-stop process with labels assigned automatically to feature vectors in order to build a set of positive and negative samples. Hence, an event model generated by using supervised learning approaches can steadily increase its accuracy. The problem of environmental impacts on asthma attacks is taken into account for evaluating the proposed method. The experimental results show that the proposed method can probably detect the prevalence of asthma risks in a specific spatio-temporal context.