4. SMP Comparison with Xerces SAX Parser


Back to the start page

In this section, we compare the performance of our projection tool with the Xerces SAX Parser. We show that for all datasets and queries discussed in our benchmark, SMP is faster in projecting the document than the Xerces SAX parser requires for merely parsing (and verifying) the document.

Xerces vs. SMP on different datasets


Xerces with SAX1/SAX2 gives the throughput of Xerces for simple parsing of the XML documents (we used the 5,000MB XMark document, the 656MB MEDLINE document, and the 685MB Protein Sequence document for this test). Avg SMP gives the the average time SMP needed for projecting the document, and is computed as the average throughput over all query tree patterns (the XMark queries, the MEDLINE queries M1-M5, and the Protein Sequence P1-P4 queries, accordingly). avg. ppl SPEX is the time needed for evaluating these queries on their documents when piping the projected document into SPEX.
Back to the start page