面向票房预测的影评情感可视分析  

Visual analysis of movie reviews sentiment for box office forecasting

作　　者：谢治海 朱敏[1] 牛红宇 梁晶 夏婷 Xie Zhihai;Zhu Min;Niu Hongyu;Liang Jing;Xia Ting(School of Computer Science,Sichuan University,Chengdu 610065,China;Sichuan Provincial Economic Information Center,Chengdu 610016,China)

机构地区：[1]四川大学计算机学院,成都610065 [2]四川省经济信息中心,成都610016

出　　处：《计算机应用研究》2020年第10期2945-2950,共6页Application Research of Computers

基　　金：四川省重点研发资助项目(18ZDYF)。

摘　　要：针对电影上映前后影评情感会发生较大变化,导致电影行业分析者分析影评情感对票房预测的影响具有一定难度的问题,提出一种基于影评情感类型与强度的自回归票房预测模型,并构建了面向票房预测的影评情感可视分析系统MRS-VIS。系统基于时空特征,提出一种空间插值可视化视图,并结合多种可视化经典视图,帮助电影行业分析者对一部电影在上映前后的影评情感进行多角度探索与分析。系统支持分析者在情感分析的基础上,通过交互操作对情感类型与强度进行编辑,进一步提高票房预测模型的准确性和可靠性。案例分析结果表明,提出的模型和构建的系统可以有效帮助电影行业分析者分析影评情感类型和修正情感。Based on auto-regressive prediction model,this paper proposed a box office forecasting model which based on the emotional types and intensity,and constructed a visual analysis system named MRS-VIS at the problem of the movie review emotion would change greatly before and after the release,which was difficult to analyze the influence of movie review emotion on box office prediction.The system based on temporal and spatial characteristics and used a variety of classic visual views,and designed a visual view based on spatial interpolation,which helped film industry analysts to explore and analyze the movie review emotions before and after the release of a movie from multiple perspectives.The system supports to edit emotion types and intensity through interactive operation based on sentiment analysis.The results of the case study show that the proposed model and the constructed system can effectively help the film industry analysts to analyze the types of review emotion and modify emotions.

关 键 词：情感分析 情感可视分析 空间差值可视化 票房预测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]