On Generating A Relational Database Schema From Collections of Json Documents

The current advancement of relational technologies for Big Data, such as NewSQL databases, has aroused the interest of the Database scientific community, as well as the industry, towards a return to the use of relational databases instead of using NoSQL technologies in some application domains. This paper presents a process based on mapping algorithms and heuristics for converting document-oriented NoSQL database schemas, or even legacy JSON datasets, to relational database schemas. The intention is to generate a data schema compatible with the schema of a set of JSON documents and with good access performance. An experimental evaluation demonstrates that our proposal is suitable to certain classes of queries.