中文摘要：

词语间语义关系是中文信息处理基础。现有算法存在语义关系抽取专业性强、依赖语料库、抽取有限等不足。通过确定符合语义关系的种子,根据种子生成谓词模板和种子模版,利用网络资源得到网页信息,将网页信息与谓词模版和种子模版进行匹配,最终得到词语间语义关系。该方法充分利用网络资源,并不受限于特定语料资源,抽取的词语关系可适用于多种语义关系的自动抽取。通过同义语义关系实验和对比实验,表明语义关系自动抽取方法具有良好的抽取结果。

英文摘要：

Semantic relations of words are the foundation of the Chinese information processing.Traditional algorithms have some limititions such as high specialization of semantic relation extraction,strong reliance on corpus,and limited extraction of semantic relations and so on.After the construction of the predicate template and the seeding template,the web information is matched with these two templates,and the semantic relations are obtained.This method does not specially depend on the corpus,and can be accommodated to multiple semantic relations.The synonymous semantic relation experiment results clemonstoate the efficiency of the proposed automatic extraction method for semantic relations.