情报杂志
情報雜誌
정보잡지
JOURNAL OF INFORMATION
2015年
6期
131-139
,共9页
况湘玲%黄光球%曹黎侠%李雪琴%罗光远
況湘玲%黃光毬%曹黎俠%李雪琴%囉光遠
황상령%황광구%조려협%리설금%라광원
复杂网络%信任网络%舆情传播%信任度
複雜網絡%信任網絡%輿情傳播%信任度
복잡망락%신임망락%여정전파%신임도
complex network%trust network%opinion propagation%trust degree
针对目前复杂信任网络较少考虑舆情传播影响信任度值的演化方向的问题,运用复杂网络方法构建了有向加权多社团复杂信任网络模型,建立了多个相异舆情态度值的传播方式模型,设计了相异舆情态度值传播至同一个节点时节点分别按照最大信任传染病模型和多数-少数信任传染病规则取舆情态度值的方法,并按照舆情传播者是否传播正确舆情态度值对传播者的信任度进行奖惩,最后对模型进行仿真并分析信任度的演化趋势。结果表明在有向加权多社团复杂信任网中进行舆情传播时,信任度趋势主要受奖惩幅度差值影响。奖惩幅度差值为正,社团和全网信任均值为上升趋势,奖惩幅度差值为负,社团和全网信任均值为下降趋势。奖惩幅度差值为零,则信任均值会出现有升有降的趋势。信任度上升的快慢则受奖惩幅度大小和社团内连接概率两个因素的影响。
針對目前複雜信任網絡較少攷慮輿情傳播影響信任度值的縯化方嚮的問題,運用複雜網絡方法構建瞭有嚮加權多社糰複雜信任網絡模型,建立瞭多箇相異輿情態度值的傳播方式模型,設計瞭相異輿情態度值傳播至同一箇節點時節點分彆按照最大信任傳染病模型和多數-少數信任傳染病規則取輿情態度值的方法,併按照輿情傳播者是否傳播正確輿情態度值對傳播者的信任度進行獎懲,最後對模型進行倣真併分析信任度的縯化趨勢。結果錶明在有嚮加權多社糰複雜信任網中進行輿情傳播時,信任度趨勢主要受獎懲幅度差值影響。獎懲幅度差值為正,社糰和全網信任均值為上升趨勢,獎懲幅度差值為負,社糰和全網信任均值為下降趨勢。獎懲幅度差值為零,則信任均值會齣現有升有降的趨勢。信任度上升的快慢則受獎懲幅度大小和社糰內連接概率兩箇因素的影響。
침대목전복잡신임망락교소고필여정전파영향신임도치적연화방향적문제,운용복잡망락방법구건료유향가권다사단복잡신임망락모형,건립료다개상이여정태도치적전파방식모형,설계료상이여정태도치전파지동일개절점시절점분별안조최대신임전염병모형화다수-소수신임전염병규칙취여정태도치적방법,병안조여정전파자시부전파정학여정태도치대전파자적신임도진행장징,최후대모형진행방진병분석신임도적연화추세。결과표명재유향가권다사단복잡신임망중진행여정전파시,신임도추세주요수장징폭도차치영향。장징폭도차치위정,사단화전망신임균치위상승추세,장징폭도차치위부,사단화전망신임균치위하강추세。장징폭도차치위령,칙신임균치회출현유승유강적추세。신임도상승적쾌만칙수장징폭도대소화사단내련접개솔량개인소적영향。
Because the evolution direction of the trust degree being affected by opinion propagation in complex trust network is seldom considered, a directed and weighted multi-community complex trust network model is built by complex network approach firstly. Secondly a model of propagation style with distinct attitude values of public opinion is established. Thirdly a method is designed for a node to select an opinion attitude value in accordance with the maximum trust value or the majority-minority trust value model when two contradictory at-titude values are spread to the node. Fourthly, the trust degrees of propagators are rewarded or punished according to whether they have spread the public opinions right or not. Finally, a simulation is carried out and the evolution trends of trust degrees are analyzed. The results show that the trends of trust degrees are mainly affected by the minus value between the reward range and punishment range. When the mi-nus value is positive, the trends of the mean trust values of the communities and of the whole net are ascending. When the minus value is negative, the trends of the mean trust values are descending. When the minus value is zero, the trends of the mean trust degrees are ups and downs. The speed of the trends is affected by the two factors of the number of minus and the linked probabilities in communities.