摘要
商品评论中含有大量的有用信息,这些信息对买方的购买行为和卖方的销售行为都有着显著的影响。商品特征作为网络评论中的关键信息,有重要的实际意义和研究价值。本文提出一种新的商品特征挖掘方法,该方法通过扩充用户词典来提升候选特征的准确性,同时引入同义词表对候选特征有效地剪枝,此外还提出情感指数的概念并以此作为从候选集中选择商品特征的依据,并从电商网站分别获取了手机和数码相机等四种商品的相关评论用于数值实验。实验结果显示该挖掘方法是可行的、有效的,不仅很好地提升现有研究结果的准确性,同时也为商品特征挖掘领域提供了新的研究思路。
Product reviews contain a lot of useful information and have a significant influence on purchasing and selling behaviors. As the critical information from online reviews, product features have important theoretical and practical values. This paper presents a new product feature mining method, which improves the accuracy of candidate sets by extending user dictionary, and introduces synonyms for effectively pruning. In addition, a concept of sentiment index is proposed to select product features from candidate sets. This study respectively crawls online reviews of 4 products like cell phone and digital camera etc. For numerical experiment, and the result shows its feasibility and effectiveness. It not only well enhances the accuracy of existing research results, but also provides new research ideas for the product feature mining area.
出处
《情报学报》
CSSCI
北大核心
2016年第1期77-83,共7页
Journal of the China Society for Scientific and Technical Information
基金
国家自然科学基金资助项目(71171030
71421001)
国家软科学研究计划项目(2013GXS2D018)
关键词
评论挖掘
商品特征
用户词典
同义词
情感指数
review mining, product features, user dictionary, synonyms, sentiment index