MetaKG: Meta-learning on Knowledge Graph for Cold-start Recommendation

Recently, knowledge graphs (KGs) are popularly used as the auxiliary information to enable more accurate, explainable, and diverse user preference recommendations. Specifically, existing KG-based recommendation methods target modeling high-order relations/dependencies from long connectivity user-item interactions hidden in KG. However, most of them ignore cold-start problems (i.e., item cold-start) analytics, which restricts their performance scenarios when involving new users or items. Inspired by success meta-learning on scarce training samples, we propose a novel based framework called MetaKG, encompasses collaborative-aware meta learner knowledge-aware learner, capture users' entities' for The aims locally aggregate preferences each learning task. In contrast, is globally generalize representation across different tasks. Guided two learners, MetaKG can effectively collaborative relations semantic representations, could be easily adapted scenarios. Extensive experiments various using three real datasets demonstrate that our presented outperforms all state-of-the-art competitors terms effectiveness, efficiency, scalability.