We reuse here the framework, the setting, and the semantic modelling for the automated synthesis of the SWS Challenge Mediator presented in the companion paper [5], and show how to exchange the synthesis paradigm from a linear-time-logic proof based algorithm to a runtime, abductive synthesis implemented in Prolog, that responds directly to situational changes. Both solutions take advantage of a policy-oriented enterprise management approach. 1 Motivation The Semantic Web Service Challenge Mediation Scenario has been solved in [5] by synthesis of its business logic within jABC [1]. There, we use semantic annotations for data and services expressed via type and activity taxonomies, and we derive automatically several versions of the business logic by means of a proof algorithm for the (semantic) linear time logic SLTL. We showed thus that, given high-level, declarative e about the goals to be achieved by the mediator, and abstract information on the available services, their purpose, an...