A Family of Bloom Filter Based A* Algorithms for Large-Scale Web Services Composition Problem

As web services become more popular, the number of available web services proliferates. As such, when there are a large number of web services available (e.g., in the range of 1,000 - 10,000), it is non-trivial to quickly find web services satisfying the given request. Furthermore, when no single web service satisfies the given request fully, one needs to “compose” multiple web services to fulfill the goal. Finding an optimal solution in such a setting is generally known as NP-complete, and thus can be doable only for a small number of web services. However, in dealing with large-scale web services composition, since the search space exponentially increases as the number of web services grows, it is important to make a wise decision on the underlying data structures and approximation algorithms. Toward this problem, in this paper, we present a family of solutions, named as BF* (BF-Star), that adopts the competitive A* graph search algorithm of AI community while utilizing the Bloom Filter as a succinct data structure. Experimental results verify the efficiency of our BF* algorithms and their suggested heuristic functions.