Optimizing the Content Generation Mechanism of Cross-Cultural Brand Communication Using Web Algorithms in Converged Media Environment

Abstract

In order to enhance the effectiveness of cross-cultural brand communication, this paper proposes the Transformer model using Self-Attention mechanism and Multi-Head Attention mechanism for content generation. Combined with the pointer network, the TP model based on the optimized Transformer model is designed. Conduct ROUGE score comparison experiments between the TP model and five mainstream copywriting generation models to evaluate their performance effects. To further validate the usability of the content generated by the TP model, this paper uses a questionnaire to analyze the audience's agreement. The results show that the performance comparison between the TP model and the AAT model, which has the best performance among the five comparison models, increases the scores in the ROUGE-1, ROUGE-2, and ROUGE-L evaluation functions by 0.00982, 0.00143, and 0.00953 in that order. The average emotional resonance identity, average value confirmation identity, and average sharing and Diffusion recognition were 3.9304, 3.9806, and 4.0202, respectively, indicating that the use of the TP model to generate marketing copy can trigger audience resonance and brand recognition, providing new ideas and methods for brand cross-cultural communication.