Ensemble Semi-supervised Entity Alignment via Cycle-Teaching

Entity alignment is to find identical entities in different knowledge graphs. Although embedding-based entity has recently achieved remarkable progress, training data insufficiency remains a critical challenge. Conventional semi-supervised methods also suffer from the incorrect newly proposed data. To resolve these issues, we design an iterative cycle-teaching framework for alignment. The key idea train multiple models (called aligners) simultaneously and let each aligner iteratively teach its successor new We propose diversity-aware selection method choose reliable aligner. conflict resolution mechanism when combining of that teacher. Besides, considering influence order, elaborately strategy arrange optimal order can maximize overall performance aligners. process break limitations model's learning capability reduce noise data, leading improved performance. Extensive experiments on benchmark datasets demonstrate effectiveness framework, which significantly outperforms state-of-the-art insufficient much noise.