Online Spatio-Temporal Rule Matching with Multiple Specifications

• Main Authors: Pei-HsuanHsieh, 謝佩璇
• Other Authors: Kun-ta Chuang
• Format: Others
• Language: en_US
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/37584068149824151788

碩士 === 國立成功大學 === 資訊工程學系 === 104 === To prevent the spreading of infectious disease in the epidemiology field, it is an important task to monitor the burst of the infectious events, which, there are lots of people get the particular disease in the close proximity in terms of both time and geography. However, not all people with infectious disease have the clear syndrome, such as the dengue fever patients. Therefore, the reported events might not reveal the actual spreading situation of the disease. The existing methods such as the dense region query, density-based clustering methods cannot directly applied to meet our needs. Therefore, to find all the suspicious infectious clusters, we propose the STMS method which is able to online detect all the specified event sets based on the user given spatio-temporal constraints. Users simply give multiple specifications which include the time interval, the circle-shaped of the spatial range constraint, the threshold of the number of the events. The STMS method is able to online detect and return the specified event sets which meet the corresponding specification. In this paper, we propose three methods which include the intuitive method, the method based on the concept of the minimum enclosing circle (MEC), and the STMS method. The synthetic data and the real data set is applied to evaluate different performance of the method in different characteristic of the data set. We find that the STMS method has the better performance in the clustering-like data; the MEC method is suited for situation of the larger number of the event threshold.