Exploiting Even Partition to Accelerate Structure Join

Among most of the approaches about XML query processing, structure join is more popular and efficient. The evaluation of structural relationship in the join, specifically the parent-child or ancestordescendant relationship, between XML element sets has imposed a great challenge on efficient XML query processing. In this paper, we present, PRIAM, an efficient algorithm for structure join by evenly partitioning the input element sets, which skips many more ancestor or descendant elements than previous algorithms. PRIAM accelerates structure join dramatically, but requires neither the order of input element sets nor any indices. Moreover, when the distribution of elements of a particular tag/label is skew, general partition-based techniques are not costefficient, however, PRIAM partitions the two input sets into different buckets evenly and only the structure join of suit buckets is helpful to the result, therefore it avoids scanning the two input sets many times and is efficient. In addition, PRIAM is portable to various coding schemes. Our extensive experiments show that PRIAM significantly outperforms previous ones.