Implementing a Personalized Recommendation System for Visual Design Using Intelligent Algorithms

In the rapidly developing Internet and e-commerce era, users’ demand for personalized visual design is increasing. Traditional recommender systems are difficult to accurately capture users’ personalized preferences, resulting in poor recommendation effect and low user satisfaction. For this reason, this study proposes a personalized recommendation system for visual design based on convolutional neural network and long and short-term memory network. First, CNN algorithm is used to extract features from visual design elements to capture complex image features; then, user behavior data is modeled by LSTM network to analyze user’s dynamic behaviors and long term preferences; finally, visual design features and user behavioral features are fused to construct a recommendation model. The experimental results show that the system has a recommendation accuracy of $42 \%$, a recall rate of $38 \%$, and a recommendation diversity of $79 \%$ among 1000 users. The results show that the recommender system based on CNN-LSTM model can effectively meet the personalized needs of users and provide high diversity of recommendations, but there is still room for improvement, indicating that the visual design recommender system combined with intelligent algorithms has a greater potential in personalized recommendation.