Efficient Service Discovery Using Social Service Network Based on Big Data Infrastructure

Service discovery and composition are a challenging issue for service computing when providing a value-added service. Existing approaches by keyword or ontology matching have limitations for locating realistic service discovery and composition that consider non-functionality or sociality. The main reason is that approaches are based on isolated services. The isolation hinders efficient discovery and composition of services. Therefore, past research suggests a social linked service network that considers relationships of functional and nonfunctional properties, and social interaction based on complex network theory, where related services can be located through sociability. However, it is difficult to create a social linked service network because services, portable devices, and sensors have been increasing in number with the progress of Big Data technology. In this paper, we propose creating a social linked service network to improve the performance of network construction by considering the distributed process on Big Data infrastructure. First, we propose an algorithm that creates a network graph using a Map-Reduce parallel programming model. Second, we evaluate the performance of network graph generation and service discovery. The experimental results show that our network created by using the Map-Reduce approach can solve the heavy computation load for the many calculations of network elements. In addition, service discovery performance is very similar to that of a none-distributed model.