A Rule-Based System for Monitoring of Microblogging Disease Reports.

Real-time microblogging messages are an interesting data source for the realization of early warning systems that track the outbreaks of epidemic diseases. Microblogging monitoring systems might be able to detect disease outbreaks in communities faster than the traditional public health services. The realization of such systems requires a message classification approach that can distinguish the messages which concern diseases from other unrelated messages. The existing machine learning classification approaches have some difficulties due to the lack of a longer history-based learning curve and the short length of the messages. In this paper, we present a demonstration of our rule-based approach for classification of disease reports. Our system is built based on the extraction of disease-related named entities. The type identification of the recognized named entities using the existing knowledge bases helps our system to classify a message as a disease report. We combine our approach with further text processing approaches like term frequency calculation. Our experimental results show that the presented approach is capable of classifying the disease report messages with acceptable precision and recall.