1 Introduction -- 1.1 Rationale -- 1.2 From Federated DBMSs to the Web -- 1.3 Overview of the Problem -- 1.4 Summary -- 2 Technological Context -- 2.1 Distributed Environments -- 2.2 Ontologies -- 2.3 Related Work -- 2.4 Summary -- 3 Architecture for Query Processing -- 3.1 A motivating example -- 3.2 Global Description of the Architecture -- 3.3 The Ontology Server -- 3.4 The Interontology Relationships Manager -- 3.5 The Query Processor -- 3.6 Summary -- 4 Accessing Underlying Data Repositories -- 4.1 Logical Schemas, Data Repositories, and Data Sources -- 4.2 Mappings: Key to Repository Heterogeneity Encapsulation -- 4.3 Main Steps in Accessing Underlying Data Repositories -- 4.4 Correlation -- 4.5 Presentation -- 4.6 Summary -- 5 Incremental Query Expansion to Multiple Ontologies -- 5.1 Integration of User and Target Ontologies -- 5.2 Plans with No Loss of Information -- 5.3 Plans with Loss of Information -- 5.4 Summary -- 6 Estimating the Loss of Information -- 6.1 Measure of Loss Based on Intensional Information -- 6.2 Measure of Loss Based on Extensional Information -- 6.3 Example: Measurement of the Extensional Loss -- 6.4 Loss for the Correlated Answer Across Ontologies -- 6.5 Translation with Loss for Roles to Be Projected -- 6.6 Summary -- 7 The OBSERVER Prototype -- 7.1 Architecture of the Prototype -- 7.2 Ontologies: Describing Domains About Bibliographic References -- 7.3 The Interontology Relationships Manager -- 7.4 The Query Processor -- 7.5 The Ontology Server -- 7.6 Testing OBSERVER: Data Retrieved for the Examples -- 7.7 Summary -- 8 Summary -- 8.1 Main Contributions -- 8.2 Features Missing in Our Approach.