Bipartite graph-based service recommendation method study

By taking the compatibility of Web service into full consideration, a BIpartite Graph based ServIce Recommendation (BIGSIR) method was proposed. BIGSIR adopted a bipartite graph to visual the Web services and the relationship between them. Based on the graph model, an effective recommendation algorithm was introduced to recommend the suitable Web services. Workflows and Web services from myExperiment were used as subjects to demonstrate the feasibility of the proposed approach. Experimental results demonstrate that apart from some isolated Web services or workflows, BIGSIR can obtain promising results that the average recommendation ranking rate of the recommended nodes is from 0.184 to 0.281. It has a better performance when compared with GRM (Global Recommendation Method). The factors that will influence the performance of BIGSIR are also explored. When the historical information is not such sufficient, the performance of our method will not be so good.The solution to address this problem was also presented.