Modeling and detection of database inference through system development life cycle to enhance data security.

Abstract

Inference is a good technique that authorizes users to deduce sensitive information from data stores which are unauthorized to their level of access. It makes confidential threat to database systems. There are three types of inference strategies: abductive, deductive and approximate. This project discusses the way of minimizing confidentiality issues in production environments which are introduced by the deductive inference techniques. Design level issues and deviation of initial design due to subsequent changes are some factors which create above problem in production databases. This project introduces a tool to detect weaknesses in database designs by validating its meta level information. This is called schema level inference detection system. This tool finds integrity constraints violations, strength of restricted associations (functional dependencies) between attributes and users accessibility to restricted associations in database using three types of rules. These rules are named as Simple constraint based rules, Attribute association based rules and Level based rules.