Abstraction in ontology-based data management

In many aspects of our society there is growing awareness and consent on the need for data-driven approaches that are resilient, transparent, and fully accountable. But in order to fulfil the promises and benefits of a data-driven society, it is necessary that the data services exposed by the organisations' information systems are well-documented, and their semantics is clearly specified. Effectively documenting data services is indeed a crucial issue for organisations, not only for governing their own data, but also for interoperation purposes. In this thesis, we propose a new approach to automatically associate formal semantic descriptions to data services, thus bringing them into compliance with the FAIR guiding principles, i.e., make data services automatically Findable, Accessible, Interoperable, and Reusable (FAIR). We base our proposal on the Ontology-based Data Management (OBDM) paradigm, where a domain ontology is used to provide a semantic layer mapped to the data sources of an organisation, thus abstracting from the technical details of the data layer implementation. The basic idea is to characterise or explain the semantics of a given data service expressed as query over the source schema in terms of a query over the ontology. Thus, the query over the ontology represents an abstraction of the given data service in terms of the domain ontology through the mapping, and, together with the elements in the vocabulary of the ontology, such abstraction forms a basis for annotating the given data service with suitable metadata expressing its semantics. We illustrate a formal framework for the task of automatically produce a semantic characterisation of a given data service expressed as a query over the source schema. The framework is based on three semantically well-founded notions, namely perfect, sound, and complete source-to-ontology rewriting, and on two associated basic computational problems, namely verification and computation. The former verifies whether a given query over the ontology is a perfect (respectively, sound, complete) source-to-ontology rewriting of a given data service expressed as a query over the source schema, whereas the latter computes one such rewriting, provided it exists. We provide an in-depth complexity analysis of these two computational problems in a very general scenario which uses languages amongst the most popular considered in the literature of managing data through an ontology. Furthermore, since we study also cases where the target query language for expressing source-to-ontology rewritings allows inequality atoms, we also investigate the problem of answering queries with inequalities over lightweight ontologies, a problem that has been rarely addressed. In another direction, we study and advocate the use of a non-monotonic target query language for expressing source-to-ontology rewritings. Last but not least, we outline a detailed related work, which illustrates how the results achieved in this thesis notably contributes to new results in the Semantic Web context, in the relational database theory, and in view-based query processing.