Probabilistic, context-sensitive, and goal-oriented service selection

  • Authors:

  • Fabio Casati;Malu Castellanos;Umesh Dayal;Ming-Chien Shan

  • Affiliations:

  • Hewlett-Packard, Palo Alto, CA;Hewlett-Packard, Palo Alto, CA;Hewlett-Packard, Palo Alto, CA;Hewlett-Packard, Palo Alto, CA

  • Venue:
  • Proceedings of the 2nd international conference on Service oriented computing
  • Year:
  • 2004

Quantified Score

Hi-index 0.01

Visualization

Abstract

In this paper we propose a novel approach and a platform for dynamic service selection in composite Web services. The problem we try to solve is that of selecting, for each composite service execution and for each step in the execution, the service that maximizes the probability of reaching a user-defined goal. We first underline the limitations of a priori approaches based on having each service provider declare non-functional parameters and on trying to select services based on some utility functions over these parameters. Then, we propose an approach that overcomes these limitations by tackling the problem a posteriori: we analyze past executions of the composite service and build, using data mining techniques, a set of context-sensitive service selection models to be applied at each stage in the composite service execution. We show the architecture of a prototype that implements this approach and we discuss its benefits over the a priori approach.