Composite Service Selection Based on Dot Pattern Mining

There are complicated correlations among atomic Web services in a composite service. Considering the influence of such correlations on the quality of composite service is important to do service selection. Now, existing selection approaches don´t consider this issue. Using the idea of division, a dot pattern based composite service selection approach is presented in this paper. The main idea of this approach is that according to the correlation degree of atomic services, dividing the composite service into a series of composite service dots; then regarding the dot as selecting unit to do service selection based on the quality of corresponding dot patterns. Experimental results show that our selection approach can improve the quality of the selected composite service.