Most Algorithms for frequent item set mining typically make the assumption that data is centralized or static. They may waste computational and I/O resources when the data is dynamic, and they impose excessive communication overhead when the data is distributed. As a result, the data mining process is harmed by slow response time. In this paper we propose a novel algorithm that uses overlapping data partitions and parallelizes the workload among machines efficiently. Experiments confirm that our algorithm results in excellent running time improvements.