陈志雄,王时绘,高榕.基于情感倾向性分析的微博意见领袖识别模型[J].计算机科学,2018,45(5):168-175
基于情感倾向性分析的微博意见领袖识别模型
Recognition Model of Microblog Opinion Leaders Based on Sentiment Orientation Analysis
投稿时间:2017-02-06  修订日期:2017-05-06
DOI:10.11896/j.issn.1002-137X.2018.05.028
中文关键词:  微博,情感倾向性,变异系数法,意见领袖
英文关键词:Microblog,Emotional tendency,Variation coefficient method,Opinion leader
基金项目:本文受国家自然科学基金青年项目(41201404),国家自然科学基金项目(41401464,11101131),湖北省科技厅面上资助
作者单位E-mail
陈志雄 湖北大学计算机与信息工程学院 武汉430062 czxpro@163.com 
王时绘 湖北大学计算机与信息工程学院 武汉430062  
高榕 武汉大学计算机学院 武汉430072  
摘要点击次数: 359
全文下载次数: 224
中文摘要:
      当前,微博意见领袖识别的研究方法纷繁多样,常见的方法有:对用户的个性化特征进行综合分析的方法和基于社交网络结构的分析方法。这些方法大多只考虑了用户的特征,未考虑用户之间的互动行为,或者未考虑 微博文本的情感因素。为此,提出了一种基于微博情感分析的微博意见领袖识别方法。首先,在基于合成情感词典的词频统计结果的基础上,利用支持向量机对微博博文进行情感分析;然后,将变异系数法用于微博属性权重的计算,以体现微博的影响力;最后,利用改进的PageRank算法在微博用户转发关系网络中预测用户影响力的扩散过程,计算用户最终影响力的大小。在新浪微博数据集上通过实验评测该方法的性能,结果表明该方法能够有效提高识别性能。
英文摘要:
      The current methods of opinion leaders are numerous.Existing methods include the comprehensive analysis method for user’s personalized features and the analysis method based on the structure of social networks.These methodsonly consider the characteristics of the users,but ignore the interaction between the users,or not take the microblog sentiment factors into consideration.Therefore,this paper proposed a microblog opinion leader identification method based on microblog emotional analysis.First of all,support vector machine is used to do emotional analysis on microblog context based on the synthetically emotional dictionary word frequency statistics results.And then the variation coefficient method is used in microblog attribute weight calculation to reflect the influence of microblog.Finally,the Page-Rank algorithm is employed to predict the diffusion process of users’ influence in the microblog,and the final influence of users is calculated.On the Sina microblog dataset,the proposed model is tested and compared with the other methods based on user attribute analysis and PageRank using the forwarding rate and coverage index.The experimental results show the superiority of proposed method.
查看全文  查看/发表评论  下载PDF阅读器