On Forgetting Relations in Relational Databases

• Autor(a) principal: Fonseca, Guilherme António Duarte
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/135417
• topic: forgetting; relational databases; relational model; General Data Protection Regulation; right to be forgotten; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática

Although not usually acknowledged as such, forgetting is a crucial aspect of human reasoning. It allows us to deal with large amounts of information, pushing irrelevant details out of our consciousness so that we can focus on the essential knowledge. Motivated by its beneficial effect on the human brain, this operation has been emulated in many formalisms in the field of Knowledge Representation and Reasoning, where several approaches to forgetting have been proposed. In common, these support computer systems dealing with inaccurate or excessive information without negatively affecting the remaining knowledge. More recently, the General Data Protection Regulation’s ‘right to be forgotten’ has given additional impetus to the study of this operation. Surprisingly, forgetting has not yet been studied in relational databases, the most widespread technology for knowledge representation. This is a serious drawback that needs to be addressed, considering the prominence of databases in our society and the relevance of the operation in numerous knowledge processing tasks. In this dissertation, we take the first steps to tackle this need, proposing a theoretical investigation of forgetting relations in relational databases. We start by introducing an alternative formalisation of the relational model, which includes a novel notion of equivalence between databases. Afterwards, we look further into the problem of forgetting. We formally define the general concept of a relation forgetting operator and present concrete operators, each aligned with a distinct view on the operation and thus with its unique features. Moreover, we illustrate the operators with examples inspired by realistic situations. Finally, we evaluate them. For that, we formalise in the form of properties the requirements that guided the definition of the operators and prove that they satisfy desirable properties. Ultimately, with this work, we motivate the importance of forgetting in relational databases and lay the foundations for its study.