The correlation that exists among data can be exploited with malicious use and it can further infer the series of sensitive information making the data accessible innocuous. An interference violation detection system is developed that protects the sensitive content of data. Semantic knowledge and schema database and is based on data dependency. Moreover, we also create a model of semantic interference  that represent the interference channels.

The inference graph is then instantiated with the SIM and it enables in query time inference violation detection. When a user poses a query the system of detection will examine queries related to his/her past log. At the same time it will calculate the inferring sensitive information. If the probability of the inference increases the pre-specified threshold then the query will be denied. The query answers can be shared among the users which can enhance the probability of the inference.

A model is developed by us that can evaluate the inference based collaborative on the sequences of queries of collaborators as well as their collaboration task sensitive levels. Studies of experiments reveal the authoritiveness of information as well as communication fidelity that is available in two key factors. It affects the achievable collaboration level.

In this paper we present a method that obstructs the users from inferring sensitive informations. In comparison to the deterministic approach in our previous works we include the relations of non-deterministic system into the interference channel. From the probabilistic data dependency we extract interference channel  and they are known as the semantic knowledge and the database schema. This is used for constructing a semantic inference model, also called SIM in abbreviated form. The possible channels of inference can be linked with the SIM from any attribute, set of the attributes of pre-assigned sensitives. The SIM attributes parematers can be computedin terms of relational table, consist of rows and columns.