Mining Sequential Patterns on a Grid-Computing Environment

This paper presents the design and implementation of a grid-computing environment for mining sequential patterns. An Apriori-like algorithm for mining sequential patterns is deployed in the proposed grid-computing environment. Apriori-like algorithm is not of very high performance in comparison to others but it is more convenient to be realized for distributed processing in a grid computing environment due to its loosely coupled processes. Two types of grids are designed, the computing grid and data grid, in the proposed environment. All grids are installed with full functions, each of which is wrapped by Globus toolkit. Grid services are invoked by the users or other grids and able to respond to the invoking side. There are 10 computers serving as grid nodes each of which is equipped with different hardware components and is distributed on two campuses. The experimental results show that the proposed grid-computing environment provides a flexible and efficient platform for mining sequential patterns from large datasets.