EAGER: The Identification of Social Systems Trust: Theory and Experimental Validation

EAGER：社会系统信任的识别：理论与实验验证

• 批准号: 1553746
• 负责人: Anna Scaglione
• 金额: $ 19.98万
• 依托单位: Arizona State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2015
• 资助国家: 美国
• 起止时间: 2015-09-01 至 2017-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1553746&HistoricalAwards=false
• 关键词: EAGER; Identification; Social; Systems; Trust

The effect of social network contacts is generally believed to be much stronger than either advertising or online reputation/rating systems. There are many theoretical models that try to predict how opinions spread among individuals that trust each other, but the experimental evidence to back them quantitatively in a real setting is still elusive. This project aims to close the gap between the theory of opinion diffusion and the several empirical studies that have been made recently on social media data. The novel idea is to exploit the presence of influential nodes that are ?stubborn?, i.e. agents who trust only themselves, to determine how opinions fluctuate in the network as a result of their activities and the relative trust among all agents in the social network. The mathematical insights from the theoretical models of opinion diffusion indicate that stubborn agents excite the social system in such a way that reverberations of their opinions in comments from others can allow one to fit appropriate system equations to this phenomenon. Another key insight is that using steady state models is a more robust method to match real data, since the fluctuations of opinions are not directly observable, only the individual actions and comments are. This effort, if successful, would be akin to realizing a ?social network Radar,? capturing a tomographic image of the hidden medium of a social group mutual trust. Preliminary results in the case of the celebrated linear De Groot's opinion diffusion model applied to Facebook data show remarkably good agreement between the graph one can extract from our method and first-hand knowledge of the group analyzed. This is a high risk high payoff project as several questions still need to be answered and tested to verify the results.

人们普遍认为社交网络联系的影响比广告或在线声誉/评级系统要强得多。有许多理论模型试图预测意见如何在相互信任的个人之间传播，但在真实环境中定量支持这些模型的实验证据仍然难以捉摸。该项目旨在缩小舆论传播理论与最近对社交媒体数据进行的多项实证研究之间的差距。这个新颖的想法是利用“顽固”有影响力的节点（即只信任自己的代理）的存在，来确定网络中的观点如何因其活动以及社交网络中所有代理之间的相对信任而波动。来自意见传播理论模型的数学见解表明，顽固的代理人以这样一种方式激发社会系统，即他们的意见在其他人的评论中产生反响，可以让人们将适当的系统方程拟合到这一现象。另一个关键见解是，使用稳态模型是匹配真实数据的更稳健的方法，因为意见的波动无法直接观察到，只能观察到个人的行为和评论。这项努力如果成功，将类似于实现“社交网络雷达”。捕捉社会群体相互信任的隐藏媒介的断层扫描图像。将著名的线性 De Groot 观点扩散模型应用于 Facebook 数据的初步结果表明，从我们的方法中提取的图表与所分析群体的第一手知识之间具有非常好的一致性。这是一个高风险高回报的项目，因为仍有几个问题需要回答和测试以验证结果。

项目成果

SUCAG: Stochastic Unbiased Curvature-aided Gradient Method for Distributed Optimization

• DOI: 10.1109/cdc.2018.8619336
• 发表时间: 2018-03
• 期刊: 2018 IEEE Conference on Decision and Control (CDC)
• 作者: Hoi-To Wai;N. Freris;A. Nedić;A. Scaglione