Privacy Preserving Record Linkage (PPRL) is widely used in data mining applications which aims to integrate data from different heterogeneous data sources while hiding the private information. In this paperwe propose a new algorithm for merging two datasets using Sorted Neighborhood Deterministic approach and a new Preservation algorithm that uses Pattern mining over dynamic queries. In contrast to the existing techniques our approach guarantees strong privacy less computational complexity and is scalable over large datasets. We provide empirical evidences to prove that our method is secure, fast and efficient than the existing methods.