Bowen Fu¹ and Carlos Gonzalez², ¹USA, ²California State Polytechnic University, USA

This study explores the use of machine learning models for optimizing social media content and predicting engagement outcomes, focusing on platforms like Xiaohongshu. The first experiment examined whether the platform’s similarity scores align with actual user engagement metrics, comparing it to alternatives like OpenAI, HyperWrite AI, and Google Gemini [1]. Results showed higher similarity scores for our platform, suggesting better alignment with engagement trends, though statistical significance was limited by sample size. The second experiment evaluated various machine learning models, including Random Forest, SVM, Logistic Regression, kNN, Decision Tree, and Isolation Forest, for classifying social media posts as "popular" or "not popular." Isolation Forest outperformed other models, demonstrating its ability to capture nuanced patterns in noisy datasets. The findings highlight the potential of AI-driven tools in improving social media content strategies while emphasizing the need for larger, more diverse datasets and advanced feature engineering for greater accuracy and scalability [2].

AI-Driven, Social Media, Optimization, Engagement