Discussion on Cluster Analysis and Data Privacy Protection Strategy of Cultural Communication Metaphor Patterns in Online Social Networks

Abstract

The existing analysis of cultural communication metaphor patterns in online social networks is limited to specific platforms, resulting in a one-sided understanding of cultural communication metaphors, and the data processing and storage process when analyzing metaphor patterns is prone to user privacy leakage. This paper introduced a cluster analysis and data privacy protection strategy to collect user-generated content from multiple online social platforms such as Weibo, WeChat, Facebook, and Douyin to ensure sample diversity. The Laplacian mechanism is used before data analysis to ensure data privacy by adding noise. Natural language processing technology is applied to use the BERT (Bidirectional Encoder Representations from Transformers) model to perform deep semantic analysis on text and identify metaphorical expressions in cultural communication. The K-means clustering algorithm is used to classify the extracted metaphor feature vectors, and different K values are set for experiments. The optimal number of clusters is determined by the elbow rule. The experimental data shows that after cluster analysis, the optimal number of clusters of the extracted feature vectors is 5, which are ”social identity,” ”emotional resonance,” ”transmission of cultural symbols,” ”expression of values” and ”cross-cultural communication,” When the optimal privacy budget is set to 0.5 during cluster analysis, the risk value reaches the minimum 0.008, and the analysis accuracy is 0.98. The data proves the effectiveness of the cluster analysis and data privacy protection strategy studied in this paper.