A Study of Inference Control Techniques

Abstract

Security is a major issue in every field and it impacts more when intruders get the information of an individual from the database either directly or indirectly. There are two approaches to break the confidentiality, either directly or indirectly. Here we study about different types of techniques which protect the confidentiality from indirect disclosure. These techniques are called Inference Control Techniques and also known as Statistical Disclosure Control methods. Indirect disclosure differs from the other security problems because in this presumptive intruders or external users deduce the information by set of available queries having low security risk i.e., they do computations on set of available information which is non sensitive and from that information they get the sensitive information. Inference control techniques protect the publicly released statistics of companies and institutions, such that presumptive user could not get any private information about any individual entity. Inference control in statistical databases is a part of information security which tries to prevent published statistical information (tables, individual records) from disclosing the contribution of specific respondents. Here we shall analyze about information loss, disclosure risk measures and performance of the various techniques. The major challenge to Statistical Disclosure Control is that modified data should be such that it provides more useful information with less disclosure risk i.e., protection should be maximized and information loss should be minimized. Since there is a trade-off between information loss and disclosure risk i.e., generally it happens that if disclosure risk is less then information loss will be more and vice versa. Here we propose some ideas which may provide optimal results.