A Framework for Automated Web Service Discovery and Selection in Service-Oriented Architectures

Service-oriented architectures often employ large scale service registries, in which manual service discovery becomes very costly and prone to human error. The current techniques frequently do not have automated processes for evaluating functional relevance and quality of service (QoS) parameters during service selection which causes high time latency and not optimal service binding. The framework presented here is one that combines semantic service annotation, ontology-based matching, and multi-criteria decision analysis in a fully automated process. The framework was tested on a registry of 4,567 public webservices, the semantic filtering was able to pruned the candidates list from 4,567 services down to 912 services, and the ranking algorithm finally retrieved the top 25 possible services based on throughput, response time, availability and reliability. Experimentally verified, they show a decrease in average discovery time from 920 ms to 640 ms and service relevance scoring from 0.62 to0.83 MRR. Results show an improvement in service discovery automation in terms of precision and speed.