摘要
准确挖掘商务网站中的用户评论对于商家进行有效的推荐具有重要意义。提出了一种新颖的用户评论抽取(CRE)算法进行评论信息的抽取。该算法采用了页面分块与信息熵的迭代计算技术实现了评论块的自动发现与抽取。实验结果证明了该算法具有较高的查全率与查准率。
Mining the customer reviews accurately in commercial websites has significant meaning in effective recommendation for trade company. A kind of novel algorithm-Customer Review Extraction (CRE) was put forward in this paper. CRE iterafively segments page and calculate the information entropy to automatically discover and extract the reviews. The experimental result has proved that the algorithm has higher recall and precision.
出处
《计算机应用》
CSCD
北大核心
2006年第10期2509-2512,共4页
journal of Computer Applications
基金
江苏省自然科学基金项目资助项目(BK2005046)
关键词
用户评论抽取
信息抽取
基于视觉的页面分块
Customer Review Extraction(CRE)
information extraction
Vision-based Page Segmentation(VIPS)