Querying semistructured data efficiently

Abstract

In this paper, we address the issue of fast query processing of semistructured data. To this end, we propose a new index scheme, called the HQ-Index. The HQ-Index consists of two indexes, the H-Index and Q-Index. The H-Index is basically a hash table built upon the path expressions. It serves as a path index. The Q-Index facilitates fast traversal to the ancestor nodes in the graph and the retrieval of requested information. Not only the H-Index and Q-Index have very simple index structures, but also the combined use of them can effectively speed up the evaluation of ordinary path queries. The results of our experiments further confirm the advantage of our approach when compared with the Dataguide, one of the most referenced index schemes for XML and semistructured data. © Springer-Verlag Berlin Heidelberg 2003.