Analysis and evaluation of document-oriented structures

Abstract

Document-oriented bases allow high flexibility in data representation which facilitates a rapid development of applications and enables many possibilities for data structuring. Unfortunately, in many cases, due to this flexibility and the absence of data modelling, the choice of a data representation is neglected by developers leading to many issues on several aspects of the document base and application quality; e.g., memory print, data redundancy, readability and maintainability. We aim at facilitating the study of data structuring alternatives and providing objective metrics to better reveal the advantages and disadvantages of a structure with respect to user needs. The main contributions of our approach are twofold. First of all, the semi-automatic generation of many suitable alternatives for data structuring given an initial UML model. Second, the automatic computation of structural metrics, allowing a comparison of the alternatives for JSON-compatible schema abstraction. These metrics reflect the complexity of the structure and are intended to be used in decision criteria for schema analysis and design process. This work capitalises on experiences with MongoDB, XML and software complexity metrics. The paper presents the schema generation and the metrics together with a validation scenario where we discuss how to use the results in a schema recommendation perspective.