Data analysis and discovery of relations between connected entity types within databases can be very labour and time intensive. The reason being that every database has its specific structure, which needs to be examined.
In this thesis, we evaluated if the reconstruction error of the matrix factorization model can be used to relate tables or entity types within a database. To test this concept, we developed a Python module that connects to a database and returns a ranked list of relations (pairs of entity types). This enables the user to identify the most informative relations and explore them further.