ARES:用于预测的情感感知自回归模型

发布时间：2018-04-18 11:33

本文选题：销量预测 + ARES　；参考：《计算机研究与发展》2013年08期

【摘要】：随着Web2.0的蓬勃发展,互联网上产生了大量由用户发表的评论,其中表达的观点看法对大众消费的影响越来越大,因此分析评论中蕴含的情感信息对产品销量的预测以及市场战略的调整有实际意义.针对这一问题,在分析图书销售领域网络评论特点的基础上,提出了相应的情感分析方法,首先利用词性列表及前缀词典完成极性词词典的自动抽取与构建,然后采用基于词典的方法对图书的评论内容进行情感分析及量化,最后通过将抽取的情感因素融合到自回归模型中,建立了新的预测模型——情感感知自回归模型(autoregressive emotion-sensitive model,ARES).实验结果表明,基于词典的分析方法能够有效地反映出评论自身的情感信息,并且融入了情感分析结果的模型在销量预测方面具有更好的准确性.
[Abstract]:With the rapid development of Web2.0, there has been a lot of comments from users on the Internet, and the opinions expressed in the comments have become more and more important to public consumption.Therefore, the analysis of the emotional information contained in the review is of practical significance to the prediction of product sales and the adjustment of market strategy.In order to solve this problem, based on the analysis of the characteristics of online reviews in the field of book sales, the corresponding affective analysis method is put forward. Firstly, the polarity word dictionary is automatically extracted and constructed by using part of speech lists and prefix dictionaries.Then, the dictionary-based approach is used to analyze and quantify the comments of books. Finally, a new predictive model, the autoregressive emotion-sensitive model, is established by fusing the extracted affective factors into the autoregressive model.The experimental results show that the dictionary-based analysis method can effectively reflect the emotional information of the commentary itself, and the model incorporating the emotional analysis results has better accuracy in predicting sales volume.
【作者单位】：大连理工大学计算机科学与技术学院;
【基金】：国家自然科学基金项目(60673039,60973068) 国家“八六三”高技术研究发展计划基金项目(2006AA01Z151) 教育部留学回国人员科研启动基金高等学校博士学科点专项科研基金项目(20090041110002)
【分类号】：TP393.09

【参考文献】