The aim of this study is to solve the problem of composing semantic Web services with AI planner. Automatic semantic Web service composition is a key technique in the research of Web service. The idea of composing semantic Web services using AI planner has been presented in several studies. In our opinion, since the different application domains and the constraint of precondition that AI planners have, none of the existing planners can deal with the problem of semantic Web services composition perfectly. In this study, we present an algorithm for translating semantic Web service composition problem to AI planning problem described in PDDL. Based on this algorithm, according to the situation of the problem, we can use the most suitable AI planner to compose Web services. One of our goals is to support the claim that AI planner can be used in the domain of semantic Web service composition. The other goal of ours is that the description of semantic Web services can be translated into a description written in PDDL. There are also several key procedures of the translating algorithm in this study.