├── Code ├── readme.md ├── Auto-regressive │ └── readme.md ├── Point │ └── readme.md ├── One-stage │ └── readme.md ├── Span-based │ └── readme.md └── Tabel-filling │ └── readme.md ├── README.md └── Data └── data.md /Code/readme.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # entity-relation-joint-extraction 2 | 汇总实体关系联合抽取方向的代码。 3 | -------------------------------------------------------------------------------- /Data/data.md: -------------------------------------------------------------------------------- 1 | NYT29 2 | 3 | 4 | 5 | NYT24 6 | 7 | https://github.com/xiangrongzeng/copy_re 8 | 9 | WebNLG 10 | 11 | https://github.com/yubowen-ph/JointER 12 | 13 | DuIE 14 | 15 | -------------------------------------------------------------------------------- /Code/Auto-regressive/readme.md: -------------------------------------------------------------------------------- 1 | | 论文 | prec | recall | F1 | 是否使用BERT | 代码链接 | 2 | | ---------- | ---- | ------ | ------------------------- | ------------ | --------------------------------------------- | 3 | | WDec | 88.1 | 76.1 | 81.7 | 否 | https://github.com/nusnlp/PtrNetDecoding4JERE | 4 | | Seq2UMTree | 79.1 | 75.1 | 77.1（认为NYT有数据偏差） | 无 | https://github.com/WindChimeRan/OpenJERE | 5 | | | | | | | | 6 | 7 | -------------------------------------------------------------------------------- /Code/Point/readme.md: -------------------------------------------------------------------------------- 1 | | 论文 | prec | recall | F1 | 是否使用BERT | 代码链接 | 2 | | ------ | ---- | ------ | ---- | ------------ | ------------------------------------------------------------ | 3 | | CasRel | 89.8 | 88.2 | 89.0 | 是 | https://github.com/weizhepei/CasRel https://github.com/longlongman/CasRel-pytorch-reimplement | 4 | | | | | | | | 5 | | | | | | | | 6 | 7 | -------------------------------------------------------------------------------- /Code/One-stage/readme.md: -------------------------------------------------------------------------------- 1 | NYT 2 | 3 | | 论文 | prec | recall | F1 | 是否使用BERT | 代码链接 | 4 | | -------------------------- | ---- | ------ | --------------------- | ------------ | ------------------------------------------------------ | 5 | | TPLinker**(TPlinkerPlus**) | | | 91.9(**92.6 (+3.0)**) | 是 | https://github.com/131250208/TPlinker-joint-extraction | 6 | | | | | | | | 7 | | | | | | | | 8 | 9 | -------------------------------------------------------------------------------- /Code/Span-based/readme.md: -------------------------------------------------------------------------------- 1 | ADE 2 | 3 | | 论文 | **NER MACRO F1** | **RE MACRO F1** | **RE+ MACRO F1** | 是否使用BERT | 代码链接 | 4 | | ---------------------- | ---------------- | --------------- | ---------------- | ------------ | -------------------------------------- | 5 | | SpERT(without overlap) | 89.25 | | 79.24 | 是 | https://github.com/markus-eberts/spert | 6 | | SpERT(overlap) | 89:28 | | 78:84 | 是 | | 7 | | | | | | | | 8 | 9 | ACE05 10 | 11 | | 论文 | **NER MICRO F1** | **RE MICRO F1** | **RE+ MICRO F1** | 是否使用BERT | 代码链接 | 12 | | ------- | ---------------- | --------------- | ---------------- | ------------ | ---------------------------------- | 13 | | DYGIE++ | 88.6 | 63.4 | | 是 | https://github.com/dwadden/dygiepp | 14 | | | | | | | | 15 | | | | | | | | 16 | 17 | -------------------------------------------------------------------------------- /Code/Tabel-filling/readme.md: -------------------------------------------------------------------------------- 1 | NYT 2 | 3 | | 论文 | prec | recall | F1 | 是否使用BERT | 代码链接 | 4 | | ------------------------------ | -------- | -------- | -------- | ------------ | ------------------------------------------------------------ | 5 | | attention as relation（SMHSA） | 88.1% | 78.5% | 83.0% | 无 | https://github.com/chenshaowei57/SMHSA | 6 | | RIN | 83.9±0.5 | 85.5±0.5 | 84.7±0.4 | 无 | https://github.com/BDBC-KG-NLP/Recurrent_Interaction_Network_EMNLP2020 | 7 | | | | | | | | 8 | 9 | 10 | 11 | ACE04 12 | 13 | | 论文 | **NER MICRO F1** | **RE MICRO F1** | **RE+ MICRO F1** | 是否使用BERT | 代码链接 | 14 | | -------------- | ---------------- | --------------- | ---------------- | ------------------ | ------------------------------------------------------------ | 15 | | multi-head | 81.6 | | 47.14 | 否 | https://github.com/bekou/multihead_joint_entity_relation_extraction | 16 | | Tabel-Sequence | 88.6 | 63.3 | 59.6 | 使用BERT预训练权重 | https://github.com/LorrinWWW/two-are-better-than-one | 17 | | | | | | | | 18 | 19 | --------------------------------------------------------------------------------