摘要
[目的/意义]随着社交网络的发展,新浪微博成为人们生活中交流的重要平台。针对微博情感的分类研究有利于舆论发现,帮助政府和企业进行舆论引导。[方法/过程]基于微博中的情感词、表情符号、否定副词、程度副词等情感知识分类算法和传统的机器学习算法,提出了一种组合分类算法。该算法解决了传统机器学习算法在情感分类时样本数据分布不均的问题;相比情感知识分类算法,该算法的分类效率大大提升。[结果/结论]实证结果表明组合分类算法在准确度、召回率、F值等方面均优于情感知识算法和机器学习算法。
[Purpose/Significance]With the development of social networks,Sina microblogging has become an important platform for people to communicate in life. In the field of microblogging,emotional analysis of micro-text is conducive to public opinion,by helping the government and enterprises to guide public opinion.[Method/Process]This paper presents a combinatorial classification algorithm based on the combination of emotional knowledge classification algorithm and machine learning algorithm based on emotional word,emoticons,negative adverbs and degree adverbs in microblogging. The algorithm solves the problem of uneven distribution of sample data in the traditional machine learning algorithm in emotion classification,and solves the problem of lowefficiency of classification of emotion knowledge classification algorithm.[Result/Conclusion]The empirical results showthat the combined classification algorithm is superior to the emotion knowledge algorithm and the machine learning algorithm in terms of accuracy,recall rate and F value.
作者
何跃
赵书朋
何黎
He Yue;Zhao Shupeng;He Li(Business School of Sichuan University,Chengdu 610065)
出处
《情报杂志》
CSSCI
北大核心
2018年第5期189-194,共6页
Journal of Intelligence
关键词
情感倾向分类
机器学习
情感知识
组合分类
emotional tendentious classification
machine learning algorithm
emotional knowledge
combined classification
