Digital Dissemination of Information Based on Knowledge Graph Recommendation Algorithm

Abstract

The rapid advancement of technology has made recommendation systems a research focus in the field of modern information processing. To enhance the accuracy of recommendation systems and effectively capture and recommend user interests in complex data environments, this study starts from the combination of knowledge graphs and graph neural networks. Firstly, the water wave network algorithm is introduced to diffuse user interest information and fully utilize the global information of the knowledge graph. Secondly, the processed knowledge graph is input into the graph neural network for deep feature learning, and finally a recommendation algorithm combining ripple network and graph neural network is proposed. The test results showed that when the final recommendation results of the model were 5, 10, 15, 20, 25, and 30, the matching degrees of the recommendation results were 99.2%, 98.3%, 97.4%, 96.5%, 95.8%, and 95.2%, respectively. On a 100% dataset, the improved model achieved a 7.52% increase in hit rate and a 14.29% increase in mean reciprocal ranking compared to the original knowledge graph convolutional network. Experiments have shown that by combining knowledge graphs and graph neural networks, the dynamic adaptability and recommendation quality of recommendation systems have been effectively improved, providing an efficient and accurate solution for information recommendation.