Towards Rule-Based Context Transformation in Mediators

Mediators are used for the integration of heterogeneous information sources. Rule{based mediators map the sources to an integrated view by transformation rules reconciling structural heterogeneity problems. But the reconciliation of semantic heterogeneity problems play a subordinate role. In this paper, we present an approach that extends rule{based mediators for the reconciliation of semantic heterogeneity problems. The discovery of the semantic heterogeneity problems requires an explicit representation also indicating the semantics of an information. In this paper, the semantic level of an information item is described by contexts. The transformation of several information sources into the integrated view includes the transformation into the context of the integrated view | the context transformation. The set of transformation rules is extended with so{called context rules for the context transformation. Context rules need an extension of the representation formalism and a new inference mechanism provided by the mediator. The declarative representation of context rules allows to propagate the context transformation into the sub-queries sent to the heterogeneous information sources and provides potential for optimization.