基于多维情感词典的B站视频弹幕倾向性分析  

Sentiment analysis of Bilibili video barrage based on multidimensional sentiment dictionary

作　　者：金丹丹 于干 JIN Dandan;YU Gan(School of Computer and Information Engineering,Fuyang Normal University,Fuyang Anhui 236041,China)

机构地区：[1]阜阳师范大学计算机与信息工程学院,安徽阜阳236037

出　　处：《阜阳师范大学学报（自然科学版）》2022年第2期99-105,共7页Journal of Fuyang Normal University:Natural Science

基　　金：安徽省高校优秀青年人才支持计划项目(gxyqZD2020054)资助。

摘　　要：针对常规情感词典不适用于B站视频弹幕领域的语料库且在情感分析上维度比较单一的问题,提出基于多维情感词典的弹幕倾向性分析。利用融合改进词林和改进知网的词语间语义相似度算法,实现不同分类维度的词典按照人的七种情感维度“乐”、“好”、“怒”、“愁”、“惊”、“恶”和“惧”进行合并,并且结合B站发布的年度弹幕、弹幕表情符号等信息设计面向B站视频弹幕领域的多维情感词典。基于自构建的多维情感词典提取视频弹幕中的情感词和情感符号,使用改进的情感值计算方法度量弹幕情感值,以可视化的形式呈现分析结果。该研究通过引入多维情感词典和改进的情感值计算方法对弹幕内容情感分析,获取电影所包含的情感信息,提供一种视频检索的方法,满足观影者个性化需求,提升视频播放平台的用户黏度。Aiming at the problem that the conventional emotion dictionary is not suitable for the corpus of the video barrage field of Bilibili and has a relatively single dimension in sentiment analysis,this paper proposes the bullet-barrage orientation analysis based on multi-dimensional sentiment dictionary.Using the algorithm of semantic similarity between words of fusion improved Cilin and improved CNKI,the dictionaries of different classification dimensions are merged according to the seven emotional dimensions of human:“joy”,“good”,“anger”,“sorrow”,“shock”“,evil”and“fear”.In addition,a multi-dimensional emotional dictionary is designed for the video barrage field of Bilibili based on information such as annual barrage and barrage emoticons released by Bilibili.Based on the self-constructed multidimensional emotion dictionary,emotive words and emotive symbols are extracted from video bullet screen,and the improved emotive value calculation method is used to measure the emotive value of bullet screen,and the analysis results are presented in the form of visualization.Through the introduction of multidimensional emotion dictionary and improved emotion value calculation method,this study analyzes the emotion of bullet screen content,obtains the emotional information contained in the movie,and provides a video retrieval method to meet the personalized needs of viewers and improve the user viscosity of the video playing platform.

关 键 词：网络爬虫 多维情感词典 可视化 文本倾向性分析 

分 类 号：TP-391[自动化与计算机技术]