Next App Prediction Based on Graph NeuralNetworks and Self-Attention Enhancement

Next mobile app prediction aims to recommend the apps that users will most likely to use next based on their historicalusage behavior. It is critical for optimizing app preloading strategies and personalized recommendations, enhancing the userexperience on mobile devices. However, it faces fundamental challenges such as interactions sparsity, rapid expansion ofthe app ecosystem and long-term interest neglect. To overcome the limitations of existing methods in next-app prediction,particularly in personalized feature extraction and temporal dynamics modeling, we propose a temporal-personalized next-app prediction framework, which employs multi-perspective graph representation learning with self-attention mechanisms toenhance user and app embeddings. It can effectively capture both long-term and short-term evolving user interests in appusage, enhancing dynamic temporal features of users and apps. Moreover, it can integrate global interactions into graphrepresentation learning by multi-perspective feature aggregations. With a context-aware attention fusion mechanism applied,we effectively integrate temporal and personalized features. The comprehensive user and app embeddings are obtained tonext-app prediction, which significantly improve the accuracy of next app prediction. Experimental results on real datasetsdemonstrate that our model outperforms other baselines.