Distantly Supervised Relation Extraction Based on Residual Attention and Self Learning
Zhiyun Zheng,Yamei Xu,Yun Liu,Xingjin Zhang,Lun Li,Dun Li
DOI: https://doi.org/10.1007/s11063-024-11497-0
IF: 2.565
2024-05-23
Neural Processing Letters
Abstract:Relation extraction is an important task in information extraction, which aims to identify the relation between two given entities. The algorithm based on distant supervision can automatically generate a large amount of annotated data, which becomes the main method to deal with the task of relation extraction. However, previous studies rely too much on the precision of supervision information and ignore the effective supervision information hidden in the case of mislabeling, which leads to the loss of supervision information. To solve this problem, we propose the distantly supervised relation extraction model based on residual attention and self-learning. The model uses residual attention to extract features, and then uses self-learning idea to generate corrected labels for training data, which are added into the training process as supervisory signals to prevent parameter error updates caused by noisy labels. The model can not only reduce the problem of mislabeling caused by distant supervision, but also makes full use of the available supervisory information in the data to improve data utilization. Experiments show that compared with the existing mainstream baseline methods, the proposed model has higher precision and recall.
computer science, artificial intelligence