An Evaluation Study of Search Algorithms for XML Streams

Abstract

Keyword-based searching services over XML streams are essential for widely  used streaming applications, such as dissemination services, sensor networks and stock market quotes. However, XML stream keyword search algorithms are  usually schema dependent and do not allow pure keyword queries. Furthermore, ranking methods are still relatively unexploited in such algorithms. This paper presents an accuracy and performance study of two keyword-based search algorithms for XML streams.Our study provides a comparison of these two algorithms by using an XPath benchmark as source of data and queries. Moreover, we also consider  a large collection of XML documents and a large set of random queries, both based on DBLP dataset. Finally, we propose a strategy that combines both algorithms and ranks the keyword-based search results.