Research on Intelligent Service Composition and Recommendation System Combining Multi-Modal Data Fusion and Deep Neural Network
Xinyue Wang · Procedia Computer Science · 2026
With the continuous development of intelligent recommendation technology, how to accurately predict user needs and provide personalized service has become a key issue in the research of intelligent service system. Due to the characteristics of high dimensional, multi-source and sparse user data, it is often difficult for traditional recommendation algorithms to effectively mine the potential needs of users when processing these complex data. Therefore, this paper proposes an intelligent service recommendation method based on multi-modal data fusion and deep neural network. First, on the premise of ensuring user privacy, unsupervised learning is used to preprocess a large number of user data, and data normalization and dimensionality reduction techniques are used to effectively alleviate data sparsity. Then, users are accurately classified based on the improved clustering algorithm, and different clustering methods are compared through the optimized evaluation index to ensure the accuracy of the clustering results. Next, combining the text and image information from users’ historical behaviors, this paper proposes a multi-modal demand prediction model based on soft attention mechanism and gated cyclic unit network. On the basis of feature fusion, the soft attention mechanism is used to effectively extract users’ deep interests, and then the time series features of users’ preferences are learned through GRU network. The experimental results show that the proposed model is superior to the traditional method in a number of evaluation indicators, including MAE, MSE and R² indicators, which increase by 5.81%, 6.45% and 6.0%, respectively, which verifies the effectiveness and application potential of this method in intelligent service recommendation. The research in this paper provides a new idea for the application of multi-modal data fusion and deep learning technology in intelligent recommendation system, and also provides technical support for improving user experience and service accuracy.