您的位置:首页 > 论文页面
商品评论的代表性信息提取方法研究
发表时间:2015-10-31 浏览量:1977 下载量:840
全部作者: | 马雪静,任明 |
作者单位: | 北京大学光华管理学院;中国人民大学信息资源管理学院 |
摘 要: | 从信息提取的角度进行聚类分析,在各类评论中提取相应数量的评论,组成代表性评论集合。在各类评论提取时,提出相似度优先和差异度优先2种代表性评论抽取方法。论文使用实际数据展示了该方法的作用和效果,并使用评价指标对2种方法进行对比。结果表明,2种方法在覆盖度上无显著差异,差异度优先的方法在冗余度上显著优于相似度优先的方法。 |
关 键 词: | 情报学;信息提取;代表性;覆盖度;冗余度 |
Title: | Extraction methods of representative information in product reviews |
Author: | MA Xuejing, REN Ming |
Organization: | Guanghua School of Management, Peking University; School of Information Resource Management, Renmin University of China |
Abstract: | This paper proposes an information extraction perspective to conduct clustering and then extract reviews from each cluster. In extracting reviews from each cluster, one can use two extract methods i.e. similar-review-first or diverse-review-first. This paper illustrates the proposed method based on reviews data and further tests on the performance of the two approaches. It is shown that the reviews set from two approaches have no significant difference on coverage, but the reviews set from diverse-review-first outperforms the one from similar-review-first on redundancy. |
Key words: | information science; information extraction; representativeness; coverage; redundancy |
发表期数: | 2015年10月第20期 |
引用格式: | 马雪静,任明. 商品评论的代表性信息提取方法研究[J]. 中国科技论文在线精品论文,2015,8(20):2183-2189. |

请您登录
暂无评论