Since XML parsing is time-consuming operation which greatly affects the performance of XML application,parallelization is an important optimization measure. Existing parsing methods need pre-parsing stage to ensure proper data partition so as to complete XML segments parsing in parallel, however, pre-parsing tends to be long-running and difficult to be optimized. This paper presented a schema which supports parallel sub-tree construction upon arbitrary XML segments. Sub-trees were merged to form whole XML tree in final stage. Experiment results indicate that our schema can efficiently realize parallel XML parsing in multi-core environment.%XML解析的高耗时特点制约着XML应用系统整体性能的提高,并行化是一种重要的优化手段.现有的并行XML解析算法存在的问题是需要通过预处理进行数据划分,才能实现分片并行完全解析处理.预处理往往很耗时,若进行优化处理,则实现复杂.提出的方法可实现对XML数据任意分片直接进行解析,并行构建各个片断中的子树,再通过子树合并获得全局的文档树.实验结果表明该方法能有效利用多核计算环境,并行实现XML解析.
展开▼
