Multi-Layer Perceptron Classifier and Paillier Encryption Scheme for Friend Recommendation System

Now a days, due to the world connectivity aspect social networking sites becomes very popular. Using this sites peoples can interact with each other; also can share the information. But on social networking sites problems of privacy as well as security arises. However, users want to connect with new friends to develop their social associations and also to get data from particular gathering of individuals. In recent days old friend recommendation technique becomes very popular so some of the online social networks (OSNs) can refer this technique. These kinds of methods may compromise the privacy. For resolving this kind of issues it require privacy-preserving friend recommendation methods for social networks. This work is motivated by requirement of friend proposition without appearing seclusion and security when using social networks. The main goal of proposed schema is to help the OSN user by safely makes trust with a more abnormal that is accomplishing by multi-hop recommendation process. For increase the online social contacts of users securely this will used. The present system makes use of secure kNN plan to achieve the secured social directed coordinating. But, due to the KNN, distance based learning is not clear and cost of calculation is very greater. To tackle this issues as well as increased the precision, designed system makes use of Multi-Layer Perceptron classifier for secure social coordinate matching. At the last, from analysis on the security as well as trial outcomes, It will show that the security, feasibility as well as precision of the designed method is higher than previous system.