A Multi-Dimensional Approach to Identifying Social Media Influencers and Personalized Marketing Strategies

In the digital era, identifying key influencers within social networks is critical for public opinion analysis, information flow control, and personalized marketing. Traditional computational methods relying predominantly on network centrality metrics often overlook crucial semantic and emotional dimensions. This paper proposes a novel computational model based on multi-dimensional data fusion, integrating community structure, user behavior, content semantics, and sentiment orientation. Leveraging advanced computer technologies such as Graph Neural Networks (GNNs), advanced topic modeling, and deep learning-based sentiment computation, we construct an end-to-end identification system. The model incorporates social network theory, participatory culture theory, cognitive dissonance theory, and social identity theory, utilizing social network analysis, multidimensional scaling, text mining, and sentiment analysis to achieve joint modeling of structural features and semantic influence. Experimental results on a Sina Weibo dataset (8,000 bloggers) demonstrate the model’s superior performance in both identification accuracy and semantic understanding compared to traditional methods—e.g., 92.3% accuracy in sentiment classification, 0.72 silhouette coefficient in user clustering, and top-50 influencer prediction F1-score of 0.88—enabling effective support for generating personalized marketing strategies.