Towards Leveraging Artificial Intelligence for NoSQL Data Modeling, Querying and Quality Characterization

In the last two decades, NoSQL databases emerged and formed an umbrella category grouping well over a hundred databases of different characteristics, and providing a new take on scalability, availability, consistency and data modeling, with the aim of conquering the classic one-size-fits-all solution represented in traditional databases. NoSQL databases' heterogeneity, flexibility and high performance allowed them to encompass the volume, velocity and variety challenges brought by the Big Data era and to fulfill the complex requirements of real-time applications. These advantages have given them a competitive edge in the database market. However, despite these added bonuses, a wider adoption of NoSQL has been hindered by a few challenges. The "schema-less" nature of NoSQL databases allowing data to be directly ingested, without defining a schema a priori, can be a deterrent for many users given its potential impact on data management and integrity, application complexity and system integration. Additionally, the heterogeneous nature of NoSQL databases creates a complex and diverse landscape of query languages and interfaces requiring extensive and wideranging expertise. In an effort to breach the standardization and democratization of NoSQL databases, I propose an approach integrating artificial intelligence techniques in NoSQL life cycles, namely: data modeling, querying and quality assessment. This approach offers multi-faceted contributions to the literature and the industry, including: a uniform design methodology for NoSQL databases, an intelligent bi-directional mapping between higher and lower level NoSQL schema representations, the generation of logical and physical NoSQL data models from requirements specified in natural language, intelligent and bi-directional generation of a NoSQL query from a prompt expressed in natural language, and a multi-stakeholder NoSQL database-agnostic quality characterization framework. With this approach, I aim to pave the way towards a fully integrated, artificially intelligent system capable of undertaking the processes of data modeling, natural language querying and quality assurance of NoSQL databases. This objective, when achieved, would not only increase the adoption of NoSQL databases by both experts and novices, but would also potentially positively impact the cost and time constraints of NoSQL database handling, as well as their evolvability, maintainability and portability.