Database Normalization as a By-product of Minimum Message Length Inference

Abstract

Abstract. Database normalization is a central part of database design in which we re-organise the data stored so as to progressively ensure that as few anomalies occur as possible upon insertions, deletions and/or modifications. Successive normalizations of a database to higher normal forms continue to reduce the potential for such anomalies. We show here that database normalization follows as a consequence (or special case, or byproduct) of the Minimum Message Length (MML) principle of machine learning and inductive inference. In other words, someone (previously) oblivious to database normalization but well-versed in MML could examine a database and -using MML considerations alone -normalise it, and even discover the notion of attribute inheritance.

Minimum Message Length in Hybrid ARMA and LSTM Model Forecasting

“…MML has been used for a variety of problems, including clustering and mixture modeling [ 29 , 30 ] ([ 19 ] Section 6.8), clustering of protein dihedral angles [ 31 ], decision graphs (as an extension of decision trees, allowing for disjunctions, or “or”) [ 32 ] (Section 7.2.4 [ 19 ]) and multi-way joins in decision graphs with dynamic attributes [ 33 ], causal Bayesian nets (or Bayesian networks, or causal nets) ([ 19 ] Section 7.4) and Bayesian nets with decision trees in their (leaf) nodes [ 34 , 35 ], inference of probabilistic finite state automata (or probabilistic finite state machines, PFSAs, PFSMs) ([ 19 ] Section 7.1) and hierarchical PFSAs [ 36 ], and (given sufficient data and time, and based to whatever degree on the above-mentioned inference of Bayesian nets) automation of database normalization [ 37 ], etc.…”

Development of a functional model for the description of protein interaction networks

“…Database normalization is the process wherein a database is transformed in order to ensure that it adheres to specific design standards that reduce data redundancy, improve data integrity, ensure the ability to perform structured queries and more importantly allow the database to be extended without the need for substantial restructuring [Date, 2002]. It is conducted mostly by humans, but it is possible to be achieved via machine learning [Dowe and Zaidi, 2010]. Typically, there are 4 levels of Database normalization (more do exist, but they are less common).…”