Data service API design for data analytics

Abstract

Over recent years, Data as a Service (DaaS) has emerged as a new paradigm in the cloud computing environment. DaaS enables data provision as a service and provides controlled access to this data through RESTful APIs. Across the entire data analytics lifecycle, data service can be regarded as a method for data retrieval and exploration. However, existing data services fall short on supporting data retrieval and exploration for data analytics, as most of them are designed to simply query data based on underlying data schema rather than being driven by data analytics. Moreover, the current data service API representations only allow analysts to make one-off queries and do not provide them with any guidance on continuously exploring data. Last but not least, current data services do not support the reuse of data exploration processes and the data derived from data analysts. Accordingly, to fill the gap discussed above, this research proposes Data Exploration as a Service (DEaaS) along with the data service architecture and RESTful API design to make data retrieval efficient and effective for data analytics. The contributions of the present research include: • A set of RESTful conversation models for depicting data retrieval patterns; • A RESTful data service architecture, incorporating an abstract data model and formalized resource design for data retrieval for data analytics; • A navigation model to facilitate the dynamic discovery and recommendation of service resources by enriching HATEOAS semantics and leveraging API call history; • An extension of the data package incorporating data processing scripts and data context information, enabling the reuse of the data exploration process and the derived data. Using a prototype implementation of DEaaS, case studies and experiments were conducted in order to carry out a comparative evaluation of the proposed data service design against a conventional data service design, named OData. The case study results show that the DEaaS approach has advantages over traditional data services in terms of usability, maturity, interoperability, discoverability, reusability, and adaptability. Moreover, the experimental results show that the proposed resource navigation approach can make DEaaS outperform existing data services in data exploration.