摘要
面对互联网上海量的在线客户评论,如何能快速有效地进行识别和选择从而发现和利用其中有用的评论,已经成为人们关注的重要问题.以体验型商品电影的在线评论为研究对象,结合文本挖掘技术和实证研究方法,从文本特征出发探索影响在线评论有用性的因素,建立在线评论有用性影响因素模型,并利用该模型对评论有用性进行分类预测.与现有相关研究相比,提出的模型总体拟合效果显著提高,并发现在线影评中积极的情感倾向、较高的正负情感混杂度、较高的主客观表达混杂度以及较长的平均各句长度,对评论的有用性具有显著的正面影响.最后预测结果表明,该模型对在线影评的有用性具有较强的判别能力.
Taking movie＇ s online customer reviews as the subject, using text mining and empirical research methods, from the viewpoint of text features of online reviews, this paper establishes a model of the online reviews impact factors and forecast reviews usefulness using this model. Compared with related researches, the explanation power of our model increases significantly, and we find that, positive attitude, high mixture of positive and negative attitudes, high mixture of subjective and objective expression forms and average sentences length have significantly positive impact on reviews usefulness. Finally, our forecasting result shows that our model has a strong power to discriminate usefulness of online movie reviews.
出处
《管理科学学报》
CSSCI
北大核心
2010年第8期78-88,96,共12页
Journal of Management Sciences in China
基金
国家自然科学基金资助项目(70771032
70890080-70890082)
关键词
口碑
在线评论
有用性
文本特征
正负情感
主客观表达形式
文本挖掘
word-of-mouth
online reviews
usefulness
text features
positive and negative sentimental attitude
subjective and objective expression
text mining
