1. INTRODUCTION
The problem of optimizing database schemata to improve performance is well-studied in relational databases. However, NoSQL databases (not based on tabular relations) often lack a formal schema specification understood by the database engine. The data model provided by the database engine is typically primitive structures with higher-level abstractions
managed within the application. Since the database engine cannot manipulate or understand this higher-level data model, it is important for the application to make ecient use of the available primitives. However, no standard tools exist for schema optimization in NoSQL databases.