Imperfect Data in Database Context. How are they Stored in Extended Relational Databases?

Abstract

Building a more accurate reality model requires taking into account imperfect information present in our knowledge and language. This paper presents several aspects of data imperfection in the database context and the appropriate frameworks for their treatment. It’s concluding that null value, possibility distribution and probability theory are the best solutions to represent incomplete, imprecise and uncertain data. For each of these problems there are some relational model extension proposals, including data representation and relational algebra.