AC-Tree: An Adaptive Structural Join Index

Kaiyang Liu

HKUST

All existing indices for structural join focus on indexing the encoding information of XML elements so as to improve the XML query performance. Consequently, such indices utilize only data characteristics, but ignore query characteristics that are important for the further improvement of the query performance. In this paper, we propose AC-tree (Adaptive Cluster-tree), which is a fully workload-aware structural join index, to provide a simple but efficient way to exploit the XML query characteristics for query performance improvement. To the best of our knowledge, AC-tree is the first structural join index to take the workload into consideration. An extensive set of experiments confirms that AC-tree outperforms competitors significantly for common XML queries.