基于多属性决策的社交网络意见领袖识别
Identification of Social Network Opinion Leaders Based on Multi-Attribute Decision Making

社交网络中意见领袖的准确定位在实际应用中具有重要意义。本文提出了一种基于社交网络的拓扑结构属性的综合多属性意见领袖识别方法。关键的拓扑结构属性包括度中心性、介数中心性、接近中心性以及K-cell中心性。通过以Karate空手道俱乐部为例进行案例分析，本研究展开了深入的讨论。在方法的具体实施过程中，首先采用选优法和满意值法对方案集进行了缩减，以提高算法效率。其次，对方案集中的未规范化属性进行了数据预处理，保证了后续分析的准确性。接下来，运用本征向量法确定了各属性的权重，为最终的综合排序提供了有力支持。最后，通过加权和法得出了方案集的整体排名。研究结果表明，本文提出的方法在社交网络意见领袖识别方面取得了显著的效果。这一方法不仅提高了识别的准确性，而且在处理复杂的社交网络结构时显示出了较强的适应性。这一研究对于深化对社交网络中意见领袖角色的理解以及提升相关应用的实际效果具有积极的指导意义。
In practical applications, the precise recognition of opinion leaders within social networks carries substantial weight. This paper proposes a comprehensive multi-attribute method for recognizing opinion leaders based on the topological properties of social networks. Key evaluation metrics include degree centrality, betweenness centrality, closeness centrality, and K-cell centrality. Through a case study using the Karate Club as an example, this study conducts an in-depth analysis. In the specific implementation of the method, the paper first employs the optimization and satisfaction value methods to reduce the solution set, enhancing algorithm efficiency. Subsequently, data preprocessing is applied to normalize non-standardized attributes within the solution set, ensuring the accuracy of subsequent analyses. Moreover, the eigenvector technique is applied to ascertain the weights associated with each attribute, providing robust support for the final comprehensive ranking. Finally, the weighted sum method is employed to derive the overall ranking of the solution set. Findings demonstrate that the method proposed in this paper is notably effective in discerning opinion leaders within social networks. This approach not only enhances identification accuracy but also demonstrates strong adaptability when dealing with complex social network structures. The findings of this research contribute positively to a more profound comprehension of the functions of opinion leaders in social networks and the improvement of practical implementations.