ODaaS: Towards the Model-Driven Engineering of Open Data Applications as Data Services

The Data-as-a-Service (DaaS, or Data Services) paradigm enables an on-demand, service-based access to data, relying on similar principles to Software-as-a-Service (SaaS). DaaS permits centralized data quality management, a uniform view and access to heterogeneous data, and enables exposing a richer, domain-specific data model to users. Within this context, we are witnessing a trend in institutions to make information public as open data. However, such information is normally released "as-is", in heterogeneous formats, requiring costly, ad-hoc pre-processing steps for cleansing and analysis of its underlying structure. This paper proposes an adaptation of the DaaS paradigm for the construction of open data applications. For this purpose, we introduce an architecture based on Model-Driven Engineering (MDE), consisting of (i) multi-level modelling for the description of domains, based on generic meta-models, (ii) a library of injectors to bring data on demand from heterogeneous sources into the MDE technical space, and (iii) a REST-based infrastructure to access the data services. This work presents the architecture of such framework and the first steps in its realization.