[目的/意义] 提出网络舆情场内观点簇丛的情感极化度测算方法，为量化舆情受众情感态势和识别极化群体提供依据。[方法/过程] 首先明确舆情受众情感极化的条件，再通过设置条件阈值筛选出满足条件的观点簇丛，在此基础上引入3个极化指标（受众吸引率、极端受众增长率、极化情感增长率）描述观点簇丛在测度时间窗口内的情感发酵程度。最后利用平滑权值，对观点簇丛在不同测度阶段的情感表现进行加权综合，得出其整体情感极化程度值。[结果/结论] 观点簇丛在每个时间窗口的情感表现能为阶段性的舆情受众情感极化干预提供判断依据，而综合的情感极化度有助于准确识别场域内的敏感话题及群体，便于网络舆情的精准管控。
[Purpose/significance] This paper proposes a method to measure the emotional polarization degree of the opinion cluster in network public opinion field, which provides a basis for quantifying the emotional situation of opinion audience, identifying strongly polarized groups.[Method/process] Firstly, the conditions of emotional polarization of public opinion audience are defined. Then, by setting conditional thresholds, the opinion cluster satisfying the conditions is screened out. Based on that, three polarization descriptive indexes(audience attraction rate/growth rate of extreme audience/growth rate of polarized emotion)are introduced to describe the emotional fermentation degree of opinion clusters in the measurement time phase. Finally, the smoothing weights are introduced to weigh and synthesize the polarization performance of each opinion cluster in the measurement time series, and the overall emotional polarization degree is obtained.[Result/conclusion] The emotional performance of opinion clusters at each time phase can provide judgement basis for phased emotional polarization intervention of public opinion audience. The comprehensive emotional polarization value is helpful to accurately identify the sensitive topics and groups in the field, and providing convenience for precise control of network public opinion.
[1] 高俊峰,宋绍成.网络舆情信息受众的观点认知距离计量研究[J].图书情报工作,2016,60(20):77-85.
[2] 乐国安,薛婷,陈浩.网络集群行为的定义和分类框架初探[J].中国人民公安大学学报(社会科学版),2010(6):99-104.
[3] MCGARTY C,BLIUC A M, THOMAS E F, et al. Collective action as the material expression of opinion-based group membership[J].Journal of social issues,2010, 65(4):839-857.
[4] YZERBYT V,DUMONT M, WIGBOLDUS D, et al. I feel for us:the impact of categorization and identification on emotions and action tendencies[J]. British journal of social psychology,2011,42(4):533-549.
[5] BKOWICZ P, SOBKOWICZ A. Dynamics of hate based networks[J].The European physical journal B,2010,83(4):633-643.
[6] 石密,刘建准.网络集体行为意向:概念、测量及形成要素[J].情报杂志,2017,36(5):101-105.
[7] 杜杨沁,霍有光,锁志海.基于分位数回归的网络群体极化度量[J].图书情报工作,2011,55(24):38-43.
[8] 张润莲,兰月新,王彩华,等.网络群体性事件演化博弈分析及对策研究[J].图书与情报, 2016(4):24-30.
[9] 黄微,宋先智,高俊峰.网络舆情场中信息受众观点群落的连接鲁棒性测度及实证研究[J].情报学报,2017,36(5):503-510.
[10] 高俊峰.网络舆情场中信息受众观点群落的凝聚鲁棒性测度研究[J].情报杂志,2018,37(4):106-113.
[11] 吴诗贤,张必兰.基于观点势场的舆情极化预测模型[J].图书情报工作, 2015,59(19):108-112.
[12] 桑斯坦.网络共和国:网络社会中的民主问题[M].黄维明,译.上海:上海人民出版社,2008.
[13] CHEN H M. Group polarization in virtual communities:the case of stock message boards[J].Ischools,2013,4(2):185-195.
[14] 查先进.信息分析[M].武汉:武汉大学出版社,2014.
[15] ZHANG H P, YU H K, XIONG D Y, et al. HHMM-based Chinese lexical analyzer ICTCLAS[C]//Second SIGHAN workshop affiliated with 41th ACL. Stroudsburg:Association for Computational Linguistics, 2003:184-187.
[16] 徐琳宏,林鸿飞,潘宇.情感词汇本体的构造[J].情报学报,2008,27(2):180-185.
[17] 高歌,罗珺玫,王宇.基于HNC理论的文本情感倾向性分析[J].数据分析与知识发现,2017,1(8):85-91.
