Fluidspread: A New Method of Maximizing Positive Influence in Online Social Networks via Fluid Dynamics
Fluidspread: A New Method of Maximizing Positive Influence in Online Social Networks via Fluid Dynamics
复制标题
DOI：
10.1109/icca51439.2020.9264488
复制
发表时间：
2020-10
期刊：
影响因子：
--
通讯作者：
Feng Wang;Jinhua She;Y. Ohyama;Min Wu
中科院分区：
文献类型：
--
作者：
Feng Wang;Jinhua She;Y. Ohyama;Min Wu
影响力最大化旨在检测在线社交网络中前k个有影响力的用户。几乎所有先前的影响力传播模型都无法同时纳入用户的态度、用户之间的互动和动态影响。然而，在这项研究中，我们通过流体动力学建立了一种新的影响力传播模型，揭示了影响力传播的时间演化过程。我们基于三个维度将影响扩散建模为流体更新过程：流体高度差、流体温度和温度差。此外，我们提出了最大化积极影响的问题，并设计了一种流体传播贪婪算法来解决它。我们对我们的方法和几个基线进行了广泛的比较，实验结果说明了 Fluidspread 模型和算法的有效性和效率。
Influence maximization aims at detecting the top-k influential users in online social networks. Almost all previous models of influence spread cannot simultaneously incorporate users’ attitudes, interactions between users, and dynamic influences. However, in this study, we established a new model of influence spread via fluid dynamics, which reveals the time-evolving process for influence spread. We modeled the spread of influence as the process of fluid update based on three dimensions: the difference of fluid height, the temperature of fluids, and the difference of temperature. Moreover, we formulated the problem of maximizing positive influence and devised a Fluid-spread greedy algorithm to solve it. We conducted extensive comparisons between our approach and several baselines, and experimental results illustrate the effectiveness and efficiency of the Fluidspread model and algorithm.