基于加权频繁子树相似度的网页评论信息抽取  

Page comments extraction based on weighted similarity of frequent sub-tree

作　　者：郝志峰[1,2] 袁琴[1] 蔡瑞初[1] 温雯[1] 骆魁永 

机构地区：[1]广东工业大学计算机学院,广州510006 [2]佛山科学技术学院,广东佛山528000

出　　处：《计算机应用研究》2017年第6期1636-1639,1658,共5页Application Research of Computers

基　　金：NSFC-广东联合基金资助项目(U1501254);广东省自然科学基金资助项目(2014A030306004;2014A030308008);广东省科技计划资助项目(2012B01010029;2013B051000076;2015B010108006;2015B010131015);广州市科技计划资助项目(2014Y2-00027);国家自然科学基金资助项目(61202269;61472089;61572143;61502108;61502109);广东特支计划资助项目(2015TQ01X140);广州市珠江科技新星资助项目(201610010101);广东省哲学社科项目(GD14XYJ24)

摘　　要：针对现有网页信息抽取方法普遍存在人工耗时大、抽取准确率低等问题,提出了一种基于加权频繁子树相似度的网页评论信息抽取方法 WTS。首先通过视觉特征对网页进行剪枝处理;然后,通过深度加权的相似度度量方法抽取最佳频繁子树;最后,通过子树对齐方法抽取评论路径并解析评论内容。通过对京东、苏宁等网站的评论内容抽取实验,验证了WTS方法比D-EEM、POL等方法在抽取产品评论信息上具有一定的优势。Aiming at the problem that existing methods are time consuming and poor performance, this paper proposed WTS method based on weighted frequent sub-tree similarity, which could extract product comments information rapidly and accurately. Firstly, the method pruned Web pages through visual characteristics. Then, extracted the best frequent sub-tree by depthweighted similarity measure. Lastly, it extracted comment path and parsed comment content. Extracting product comments from the wcbsite of JD, SN and so on, experimental results show that WTS has better performance in extracting product comments than D-EEM and POL.

关 键 词：产品评论 抽取 加权频繁子树 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]