Online-Review Oriented Tags Extraction for Product Features

摘要：

　　Nowadays, hundreds of thousands of reviews are given to a product in e-commerce platforms.In order to access to the helpful information from the vast resource, a tag extraction approach based on dependency parsing patterns has been proposed in the paper.The crucial problem of the extraction is to identify the relationships between features words and the corresponding opinion words from reviews.Sentiment lexicons have been adapted to detect the opinion words, while associations between features and opinions have been identified by employing the syntactical dependency parsing and a set of filtering processing.The method has been tested in a real mobile phone-specific comments dataset;the performance (Fl-score) of the system can achieve to nearly 60 % for Chinese text specifically, much better than the compared similar approach.The resultindicates the effectiveness and feasibility of the method for the feature tag extraction task.

关键词： reviews feature extraction dependency Parsing

作者: NIE Hui DU Jiazhong WU Yijun

作者单位: School of Information Management, Sun Yat-sen University, Guangzhou 510275, China

会议类型: 国际会议

会议名称: 第一届信息获取与知识服务国际会议暨第六届搜索行为与用户认知研讨会

会议地点: 武汉

会议语种:英文

页码: 95-99

在线出版日期: 2014-10-10（万方平台首次上网日期，不代表论文的发表时间）