In real world as dependence on World Wide Web applications increasing day by day they transformed vulnerable to security attacks. Out of all the different attacks the SQL Injection Attacks are the most common. In this paper we propose SQL injection vulnerability prevention by decision tree classification technique. The proposed model make use famous decision tree classification model to prevent the SQL injection attacks. The proposed model will filter the sent HTTP request by using a decision tree classification based attack signatures. We test our proposed model on synthetic data which given satisfactory results.

No abstract

The motivation of Privacy Preserving Data Mining (PPDM) is to obtain valid data mining results without access to the original sensitive information. The different privacy preserving technique on Perturbation based PPDM approach introduces random perturbation to individual values to preserve privacy before data are published. This proposed work is based on perturbation based privacy preserving data mining. Here random perturbation approach is applied to provide privacy on the data set. Previously privacy is limited to single level trust in providing privacy to the data but now it is enhanced to multi level trust. The problem with existing multi level trust PPDM algorithms is that they fail to protect form non linear attacks. Considering that this proposed work make uses enhanced batch generation to provide privacy in the multi level trust in which data will perturb multiple times so that it can avoid non linear attacks.

Entity resolution is the problem of recognizing which entry in database refers to same cluster.in this we have to run the ER in order to reduce the running time and to obtain good results. This paper investigates how we can reduce the running of ER with minimum amount of work using k-means clustering algorithm. In this, clustering can be done according to the matching of entries. We introduce a concept of technique called as k-means clustering to maximize the matching of entries identified using a limited amount of work. We illustrate the potential gains of this entity resolution approach using k-means.