Uma álgebra ER para consultas em bancos de dados NoSQL: implementação de operadores adicionais e análise de desempenho

• Ano de defesa: 2020
• Autor(a) principal: Cabral, João Vitor Lopes
• Orientador(a): Ciferri, Ricardo Rodrigues
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação - PPGCC
• Palavras-chave em Português: NoSQL; Bancos de dados; Consultas; Entidade Relacionamento; Mapeamento
• Palavras-chave em Inglês: MongoDB; Databases; Queries; Entity Relationship; Mapping
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAO
• Link de acesso: https://repositorio.ufscar.br/handle/ufscar/13159

Databases and database systems are essential to modern life. In a scenario of increasing data handling arose the need to find newer ways to store and process data that are able to cope with the rapid evolution of society. NoSQL databases are an alternative to this scenario but the lack of standard between each database implementation increases the difficulty to perform some tasks when compared to a SQL database. To solve this problem, this paper increments the work of Noguera e Lucrédio (2019), using the query language it improves the join operation and implements the operations Cartesian Product, Selection and Projection that were proposed by Parent e Spaccapietra (1984). In order to make the metamodel creation process more friendly, a textual representation was created for the metamodels and a parser for generating code compatible with the algorithm. Two software systems were analyzed to validate the MongoDB code generated by the algorithm and to check if the query result conforms to the structure defined by the algebra. The query performance was also analyzed, comparing it to the performance of queries that were handcrafted by a Software Engineer. This paper shows that the ER algebra is compatible with document-oriented NoSQL databases and that automated query generation does not significantly affects performance.