摘要
针对结构化网页中人物实体全局模式构建问题,提出了基于SVM的动态构建方法。构建分为两个阶段,第一个阶段是来自同一数据源的人物实体结构化实例到人物实体局部模式的转化,第二个阶段是利用SVM分类器完成人物实体局部模式到人物实体全局模式的映射。本方法能适应数据源的不断变化,保证了全局模式的完整性。通过实验,验证了构建算法的有效性和可行性,并对随着结构化网页不断增多时全局模式的稳定性进行了考察。
To constructed person entity global schema in structured web pages,a method of dynamic construction with SVM is proposed, it has two stages. The first stage is to transform person entity in- stances from the same data source to person entity local schema. The second stage is to map person entity local schema to person entity global schema with the SVM classifier. Our method can adapt to the change of data sources and ensure the completeness of the global model. Through experiments, the effectiveness and feasibility of the construction algorithm are verified and the stability of the global schema with increasing of structured web pages is studied.
出处
《计算机工程与科学》
CSCD
北大核心
2014年第10期1888-1893,共6页
Computer Engineering & Science
