Ordered Solution Generation for Implicit AND/OR Search Spaces
In this paper we address the problem of generating alternative solutions for implicit AND/OR DAGs which are described using a start state and a set of transformation rules. Unlike the explicit representation of AND/OR graph, for implicitly specified AND/OR graphs, the transformation rules are used to dynamically make a portion of an AND/OR graph explicit as per requirement. We propose an algorithm, ASGAO*, for generating alternative solutions for implicitly specified AND/OR graphs. Experimental results in several domains show that, on an average, our proposed algorithm performs hundred times better with respect to running time and fifty times better with respect to space than the existing algorithm. In view of the renewed research interest towards applying AND/OR graphs in domains like graphical model, constraint satisfaction, and service composition, it is not hard to see useful applications of ASGAO*.
KeywordsService Composition Transformation Rule Terminal Node Constraint Satisfaction Heuristic Function
- 1.Russell, S., Norvig, P.: Planning. In: Artificial Intelligence: A Modern Approach, 2nd edn., pp. 375–461. Prentice-Hall, Englewood Cliffs (2003)Google Scholar
- 3.Lang, Q.A., Su, Y.: AND/OR graph and search algorithm for discovering composite web services. International Journal of Web Services Research 2(4), 46–64 (2005)Google Scholar
- 5.Elliott, P.: Extracting the k best solutions from a valued AND/OR acyclic graph. Master’s thesis, Massachusetts Institute of Technology (2007)Google Scholar
- 6.Martelli, A., Montanari, U.: Additive AND/OR graphs. In: Proceedings of the 3rd International Joint Conference on Artificial Intelligence, Morgan Kaufmann Publishers Inc., San Francisco (1973)Google Scholar