UCLAO* and BHUC: Two Novel Planning Algorithms for Uncertain Web Service Composition

Abstract

The inherent uncertainty of Web service is the most important characteristic due to its deployment and invocation within a real and highly dynamic Internet environment. Web service composition with uncertainty (U-WSC) has become an important research issue in service computing. Although some research has been done on U-WSC via non-deterministic planning in Artificial Intelligence, they cannot handle the situation that uncertain Web services with the same functionality exist in a service repository and could not get all of possible solution plans that constitute an uncertain composition solution for a given request. To solve above research challenges, this paper models a U-WSC problem into a U-WSC planning problem. Accordingly, two novel uncertain planning algorithms using heuristic search called UCLAO* and BHUC, are presented to solve the U-WSC planning problem with state space reduction, which leads to high efficiency of finding a service composition solution. We have conducted empirical experiments based on a running example in e-commerce application as well as our large-scale simulated datasets. The experimental results demonstrate that our proposed algorithms outperform the state-of-the-art non-deterministic planning algorithms in terms of effectiveness, efficiency and scalability.