As information systems are increasing in scope, they depend on many diverse, heterogeneous resources. These resources are typically developed and maintained autonomously of most of the applications that use their results. While immediate applications tend to be inventory control, payroll, production control and the like, eventually the data become important to support high-level application, as planning and decision-making. Decision support applications are typically designed subsequently and independently. Planning support is synchronized with management objectives, and has to rely on existing sources, since it is rare that sufficient time is available to build planning systems and their data collection from scratch. Other sources of information are systems as digital libraries, geographic information systems, and simulations. Dealing with many, diverse, and heterogeneous sources overwhelms high-level applications with excessive emphasis on irrelevant, but crucial details. Mediators provide intermediary services, linking data resources and application programs. Their function is to provide integrated information, without the need to integrate the data resources. Specifically, the tasks required to carry out these functions are comprised of 1. Accessing and retrieving relevant data from multiple heterogeneous resources 2. Abstracting and transforming retrieved data so that they can be integrated 3. Integrating the homogenized data according to matching key data 4. Processing the integrated data to increase the information density in the result. This processing adds value by converting data to information. Mediator modules provide intermediary services in information systems, linking data resources and application programs. Programs that led to the concept of mediation were either constructed to support specific applications, or extended services from databases. Mediators are being built now by careful domain knowledge acquisition and hand crafting the required code. In this presentation we present the conceptual underpinning for automating the mediation process. Our goal is not fully automatic code generation, since programming knowledge is necessary to provide the added value that offsets the costs of having the extra architectural layer due to formal mediation. The concept is based on the extraction of a hierarchical domain model out of the general network representing the available resources. Associated with the method are domain ontologies. Ontologies list the terms used by the models, and document their relationships. These terms provide the semantic foundation needed to perform the match operation. This material is prescriptive and forward-looking, rather than a research report. The objective of presnting it now is to gain conceptual coherence among the projects that have adopted a multi-layer information architecture, while the implementations can differ greatly. We will make references to some current work and applications.