Distantly Supervised Named Entity Recognition with Self-Adaptive Label Correction

Named entity recognition has achieved remarkable success on benchmarks with high-quality manual annotations. Such annotations are labor-intensive and time-consuming, thus unavailable in real-world scenarios. An emerging interest is to generate low-cost but noisy labels via distant supervision, hence label learning algorithms demand. In this paper, a unified self-adaptive framework termed Self-Adaptive Label cOrrection (SALO) proposed. SALO adaptively performs correction process, both an implicit explicit manners, turning into correct ones, benefiting model training. The experimental results four benchmark datasets demonstrated the superiority of over state-of-the-art distantly supervised methods. Moreover, better version by ensembling several semantic matching methods was built. Experiments were carried out consistent improvements observed, validating generalization proposed SALO.