Optimizations for query index updating in finite automaton based XML stream filtering systems

XML stream filtering is one of the most popular research topics in XML research area. Many XML stream filtering systems are based on finite automaton (FA). This kind of systems proves to have high performance and scalability in matching the XML-encoded stream to large numbers of queries. The filtering engine is the most important component of a stream filtering system. Single updating and bulk updating approaches for the query index of the filtering engine have been studied in previous works. In this paper, we optimize the previous updating techniques and propose to actually delete useless states immediately after updating the query index to improve the filtering performance. We also design a hybrid query index structure to perform the insertions and deletions of the new arriving queries simultaneously to further reduce the updating cost. Our preliminary experiments show that our approaches provide significantly better scalability and updating performance when compared with existing approaches. Pruning useless states can improve the filtering performance as well.