Exploring Distinctive Features in Distant Supervision for Relation Extraction

Learning the Distinctive Pattern Space Features for Relation Extraction

我们提供NYT数据集，该数据集一共包含233081实体对，由FreeBase对齐，关系数量为57（如果使用53关系的可自行过滤多余的关系及句子）。

Distant Supervised Relation Extraction with Cost-Sensitive Loss

远程监督

DavidG.Lowe 2004年关于SIFT的更新journalpaper，demo及其过往信息，可参考http://www.cs.ubc.ca/~lowe/keypoints/

Inrelationextractionforknowledge-basedquestionanswering,searchingfromoneentitytoanotherentityviaasin

NLP中的关系抽取的文献综述，内容丰富全面，值得学习。

Deep Ranking Based Cost-sensitive Multi-label Learning for Distant Supervision Relation Extraction

Adversarialtrainingformulti-contextjointentityandrelationextraction

Pstage_03_KLUE_Relation_extraction 训练 python train.py 推理 python inference.py --model_dir = [model_pa