New Query Abstraction Methods for Ontology-Based Data Access

Ontology-Based Data Access (OBDA) systems integrate multiple data sources by mapping them to a central ontology. A key challenge in OBDA is query abstraction, which involves translating data-level queries into a more abstract form at the ontology layer. This process is crucial for efficient and consistent data retrieval across disparate sources. The research addresses the complexities of achieving perfect query abstractions, where a direct, exact translation might not always be feasible. To overcome this, the concepts of minimally complete and maximally sound abstractions have been developed, aiming to provide the best possible approximations when perfect abstraction is elusive. The paper introduces an extension to Union of Conjunctive Queries (UCQs) that incorporates limited inequality and a special predicate for database constants. This extension, while not increasing computational complexity, enables the expression of minimally complete abstractions and, consequently, perfect abstractions when they exist. The study also establishes a novel link between maximally sound abstractions and the notion of maximum recovery from data exchange, offering new theoretical insights and practical approaches for OBDA systems.