Automated Composition of Web Services by Planning at the Knowledge Level

Abstract

In this paper, we address the problem of the auto-mated composition of web services by planning on their “knowledge level ” models. We start from de-scriptions of web services in standard process mod-eling and execution languages, like BPEL4WS, and automatically translate them into a planning do-main that models the interactions among services at the knowledge level. This allows us to avoid the explosion of the search space due to the usu-ally large and possibly infinite ranges of data val-ues that are exchanged among services, and thus to scale up the applicability of state-of-the-art tech-niques for the automated composition of web ser-vices. We present the theoretical framework, im-plement it, and provide an experimental evaluation that shows the practical advantage of our approach w.r.t. techniques that are not based on a knowledge-level representation. 1