Semantic fusion: A deep hybrid multimedia recommendation system harnessing textual and visual features

The increased variety of content in e-commerce and entertainment services creates a new research gap in the field of personalized recommendation systems. This study introduces a personalized recommendation system based on Deep Hybrid Semantic Multimedia Recommendation Systems (D-HSMR), aimed at improving the relevance and effectiveness of Multimedia Recommendation Systems (MMRS). Traditional recommendation systems are incapable of dealing with a variety of data because the nature of the data is unstructured and heterogeneous, which causes cold-start and data sparsity problems and can weaken the performance of the model. This study introduces multimedia recommendation algorithms using a hybrid model based on the fusion of textual and visual semantic features, which can effectively achieve personalized recommendations. Meanwhile, this study uses deep learning to improve the accuracy of the recommendations by incorporating textual and visual features using the BERT and RCNN algorithms, respectively, to further reduce the computational burden on the system. Finally, to verify the performance of the model, statistical analysis was performed and compared with other state-of-the-art baseline models. The evaluation of Google Play and Amazon datasets demonstrates significant accuracy improvements: up to 94% for Google Play and 77% for Amazon, surpassing state-of-the-art methods. Therefore, the proposed personalized recommendation model, utilizing an optimized Deep Hybrid Semantic Approach, is effective in enhancing multimedia recommendation systems.