Classifying XML Data of Semantic Sensor Networks

Abstract

Semantic Web technology will bring great benefits to sensor networks because it is a powerful tool to deal with large-scale, complex and heterogeneous networks. However, application of semantic technology in WSN is only a vision until the semantic processing of sensor data is achieved. In semantic Web, XML will become the main data transmission format. Therefore, classifying XML data in sensor Web is a crucial issue for introducing semantic technology into WSN. Moreover, classification algorithm for XML data is the foundation of utilizing and mining data of sensor Web. Unfortunately, nearly no algorithms or tools can be directly used for this purpose. To fill this gap, a novel classification method of XML data in semantic sensor Web based on Boosting methodology is proposed in this article. The novel method can cope with variants conditions such as dynamic XML documents stream, lack of labeled samples and incremental learning. A system is implemented to apply the novel method in a real-world sensor network. The performance of the algorithm is evaluated including its accuracy, time consumption and robustness. Experimental results demonstrate that the novel method presented in this paper can achieve high precision with a low computational overhead. Moreover, it is a general and reliable algorithm.