Semantic-Based Grid Workflow Intelligent Re-Use

Grid Workflow combines the areas of Workflow Management System and Grid technology, providing a very promising solution for distributed collaboration systems. As a fundamental infrastructure of Grid environment, Grid Workflow is not only a tool or a problem-solving method, but also a special type of re-usable knowledge. Therefore, the sharing and intelligent re-use of Grid Workflows on a large-sale is a critical problem which needs immediate focus, as we claim the ultimate goal of Grid is an integrated intelligent environment for problem-solving.Based on the Planning Theory, this paper articulates and implements the Semantic-based Grid Workflow Intelligent Re-use framework. The process to share and re-use workflows, is first to express and store in a unified way the massively existing distributed and heterogeneous models as knowledge, then use these knowledge to build or to help build new solutions when new problems arise. Workflow Re-use can be divided into two categories: 1) Re-use of single model to solve similar case through model search and modification; 2) re-use of multiple models to solve new complicated problem through model combination. The purpose is to simplify the Grid applications, to reduce waste upon re-modeling over similar problems, and finally, to improve the ability of Grid as a global collaborative environment.This framework is comprised of two kinds of information description formats—Workflow Semantics and ECA-based logic expression, as well as four core modules—Re-use module which includes semantic inquiry and Planning algorithm, Semantic Configuration, Workflow Slicing, and Model Adapter, in support of complete workflow modeling and re-use cycle.The Semantic Description Structure has the feature of object-oriented inheritance, scalability and configurability, supporting the first mechanism of re-use. The unified ECA-based Workflow Expression and Model Adapter are introduced to support transformation between different model patterns in order to make full use of existing workflows. The Planning algorithm is implemented for workflow combination to realize the second mechanism of re-use. Finally, this paper introduces the concept of Workflow Slicing Technique, to recognize and extract the highly coherent and reusable workflow partitions within models, so that the resource of re-use can be dramatically enriched.