The increasingly complicated workflow systems necessitates the development of automated workflow recommendation techniques, which are able to not only speed up the workflow construction process, but also reduce the errors that are possibly made. The existing workflow recommendation systems are quite limited in that they cannot produce a correct recommendation of the next node if the upstream nodes/sub-paths that determine the occurrence of this node are not immediately connected with it.
To solve this drawback, we propose in this paper a new workflow recommendation technique, called FlowRecommender. FlowRecommender features a more robust exploration capability to identify the upstream dependency patterns that are essential to the accuracy of workflow recommendation. These patterns are properly register offline to ensure a highly efficient online workflow recommendation.
The experimental results confirm the promising effectiveness and efficiency of FlowRecommender.
|Cite as: Zhang, J., Liu, Q. and Xu, K. (2009). FlowRecommender: A Workflow Recommendation Technique for Process Provenance. In Proc. Australasian Data Mining Conference (AusDM'09) Melbourne, Australia. CRPIT, 101. Kennedy P. J., Ong K. and Christen P. Eds., ACS. 55-62 |
(local if available)