Efficient LCA based keyword search in XML data

  • Authors:

  • Yu Xu;Yannis Papakonstantinou

  • Affiliations:

  • Teradata, San Diego, CA;University of California, San Diego, CA

  • Venue:
  • EDBT '08 Proceedings of the 11th international conference on Extending database technology: Advances in database technology
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Keyword search in XML documents based on the notion of lowest common ancestors (LCAs) and modifications of it has recently gained research interest [10, 14, 22]. In this paper we propose an efficient algorithm called Indexed Stack to find answers to keyword queries based on XRank's semantics to LCA [10]. The complexity of the Indexed Stack algorithm is O(kd|S1| log |S|) where k is the number of keywords in the query, d is the depth of the tree and |S1| (|S|) is the occurrence of the least (most) frequent keyword in the query. In comparison, the best worst case complexity of the core algorithms in [10] is O(kd|S|). We analytically and experimentally evaluate the Indexed Stack algorithm and the two core algorithms in [10]. The results show that the Indexed Stack algorithm outperforms in terms of both CPU and I/O costs other algorithms by orders of magnitude when the query contains at least one low frequency keyword along with high frequency keywords. This is important in practice since the frequencies of keywords typically vary significantly.