Mining module for adaptive XML path indexing

Recent proposals of adaptive path indices utilize mining algorithms to find frequently used paths from the query workload. However, the interaction of mining and indexing is not well established yet. In this paper we propose a new mining module for adaptive path indexing, which provides a scheme for minimal support correction based on the Borders algorithm. We illustrate the method with a new class of mining patterns which is suitable for Web based queries. We have implemented the scheme as part of the APEX index and we present an experimental evaluation of it.