Research on personalized clothing recommendation system based on AIGC

The purpose of this paper is to explore the effectiveness of Artificial Intelligence Generated Content (AIGC) technology in multimodal personalized clothing recommendation system. The features of image and text are extracted by graph neural network, and then the visual information of fashion item is extracted by using convolutional neural network (CNN) model to construct a graph neural network based on the relationship between fashion item, clothing, and user. The interactive learning relationship between user-user graph and clothing-garment graph is clarified by adjacency matrix normalization and improved GRCN technique, and finally a personalized clothing recommendation model with fused latent representation is proposed. The results show that the features of higher order neighbor nodes can be encoded by fusing multimodal convolutional neural networks, which solves the problems of low efficiency of message transfer between long-distance nodes and loss of global information, and indicates that the multimodal personalized clothing recommendation model proposed in this paper has good performance. The recommendation effect of the clothing recommendation model proposed in this paper is much higher than that of random recommendation, and the mean values of the examiner’s feedback on the ratings under the two recommendation modes are 8.54 points and 5.83 points, respectively. In addition, the prediction accuracy, sorting accuracy (96.62%) and coverage (>90%) of this paper’s model recommendation are good, proving that the recommendation results of this paper’s recommendation model are reasonable and effective.