一种高效稳定的链接分析算法  

Efficient and stable link analysis algorithm

作　　者：马慧芳[1] 史忠植[2] 

机构地区：[1]西北师范大学数学与信息科学学院计算机系,兰州730070 [2]中国科学院计算技术研究所智能信息处理重点实验室,北京100190

出　　处：《计算机应用研究》2011年第2期488-491,共4页Application Research of Computers

基　　金：西北师范大学青年教师科研能力提升计划项目(NWNU-LKQN-10-1)

摘　　要：互联网中网页的快速增长使得增量链接分析算法成为网页排序的一种有效手段,可以避免因重复建模未曾变化的网页而带来的时间和资源上的浪费。链接变化规律的发现和利用是增量链接分析技术的一个关键。提出了一种高效稳定的改进PHITS模型——增量PHITS(IPHITS)。该模型充分应用Folding in机制合理利用已有训练结果,从而将新增文本及链接信息高效地融入已训练好的PHITS模型,有效地克服了PHITS模型无法适应动态变化的链接结构以及不稳定的问题。在链接网页和互引文献的互连数据上的实验表明,IPHITS具有很高的时间效率和更好的稳定性。With massive and ever increasing of pages in the Web,incremental link analysis algorithm has become a promising method to achieve on-line information ranking.Its main advantage is the resource economization which comes from the avoidance of modeling unchanged pages.The key for incremental link analysis is to find out how pages change.This paper introduced an improved efficient and stable PHITS model,which was referred to as IPHITS.The model took advantage of Folding in technique to make full use of the existing training results and thus integrates the newly arriving documents and links into the original model in an efficient way.Another obvious advantage of the model was that it could adapt the dynamic changes of document structure stably.Experimental results on hyperlinked Webpages and cited papers demonstrate that IPHITS is more efficient and stable than that of the nave one.

关 键 词：链接分析 稳定性 PHITS 增量PHITS 增量算法 结构挖掘 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]