XML Index Query Technique Based On Path Expression

How to retrieve the information that people need from vast amounts of XML data is a hot issue in academic research currently. Structural connection, as the core operation of XML query, has a large number of research results in terms of how to improve structural join algorithm. To improve query performance, the researchers introduced the index techniques. The current structural join algorithms and index techniques still have some problems, in which there is room for improvement. At present, the structural join algorithm that requires frequent disk I/O operations, and XML index techniques have too large storage space to fully utilize their own advantages in practice. How to take better ways to achieve a balance between performance and footprint size, which contradict with each other, is worth studying.Developing the study from the problem mentioned above, this paper creates index to reduce access times of the structural connection operation to disk. In the course of inquiring, in terms of excessive containing operation, useless connection is reduced to enhance the inquiry efficiency. This paper mainly includes the following two aspects:First, this paper presents a new index structure, in which the label path is introduced into the concept of the label stream, the label path and the XML path are integrated in the Path Index, and in the containing operation, with the operation to return to the original position of the label path only and to pass once magnetic disc I/O only, a set of target node can come into being. Meanwhile, bitmap is introduced and the cost of the bitmap update is analyzed and compared. The index requires less storage space, and has good performance in query.Secondly, most of current indexes can not handle the path expression containing the predicate, and keyword query does not consider the path information XML. This paper presents a Keyword query methods based on semantic, in which, when the label expression containing predicate is dealt with, the index is established for the same label version, and path matching and keyword matching are combined together in order to enhance query efficiency.