基于情感分析的网络谣言识别方法

发布时间：2018-05-05 19:23

本文选题：情感分析 + 情感词典　；参考：《数据分析与知识发现》2017年07期

【摘要】：【目的】提出一种基于情感分析技术自动识别特定领域谣言的方法。【方法】界定高、低质量信息源,在假设高质量信息源信息更可靠的情况下,通过基于情感词典的情感分析方法,量化高质量信息源与低质量信息源对特定对象的情感差异,判定低质量信息源提供的信息是否属于谣言。【结果】将该方法应用于"食品养生"、"医学健康"两个领域进行谣言识别。在30个疑似谣言案例中准确识别出23个谣言案例,准确率为76.67%。本文提出的谣言识别方法在谣言预测方面的F值为83.34%,查全率为71.42%,查准率为100%;在非谣言文本预测上的F值为72.73%,查全率为100%,查准率为57.14%。【局限】未实现不同信息源数据自动抽取,每个谣言案例下的人工收集的谣言数量有限。【结论】本文基于情感分析的谣言识别方法对特定类型的谣言是有效的。
[Abstract]:[objective] to propose a method for automatically identifying rumors in a specific domain based on affective analysis techniques. [methods] to define high and low quality information sources, assuming that high quality information sources are more reliable, The emotion analysis method based on emotion dictionary is used to quantify the emotional difference between high quality information source and low quality information source. [results] the method was applied to "food health" and "medical health" to identify rumors. Of the 30 suspected rumor cases, 23 were accurately identified, and the accuracy was 76.67. The rumour recognition method proposed in this paper has a F value of 83.34, a recall rate of 71.42 and a precision of 100, and a F value of 72.73, a recall rate of 100 and a precision of 57.14.The [limitation] does not realize automatic extraction of data from different information sources. In each case, the number of rumors collected manually is limited. [conclusion] this method based on emotional analysis is effective for certain types of rumors.
【作者单位】：中山大学资讯管理学院;
【基金】：国家社会科学基金项目“用户评论情感分析及其在竞争情报服务中的应用研究”(项目编号:11CTQ022) 广东省科技专项“基于内容的科技文献分析服务平台”(项目编号:2016B030303003)的研究成果之一
【分类号】：G206;TP391.1

【相似文献】