Theoretical Background

Mmcat is a framework for modeling multi-model data. It is based on category theory which is abstract enough to contain all currently used data models. The necessary theoretical background is described in the article¹. The brief summary of the most important concepts is here.

The generality of the theory allows us to create a category which describe multiple logical models from different databases at once. Thus, we are able to model any combination of different logical schemas with one category. This object is called schema category.

The next step is to use the algorithms the framework provides to import data from all the databases to a unified representation called instance category. There are also algorithms which can export the data from the instance category to specific databases.

The advantage of this approach is this - we can convert data from any database A to any other database B simply by transforming them from A to the instance category and from there to B. The only thing we need is to create a wrapper for each involved database system that would handle the platform specific requirements of given database.

Each logical model is mapped to the schema category via its own mapping. This structure defines both the mapping and the logical model itself.