search

nanguoshun / LSR

Pytorch Implementation of our ACL 2020 Paper "Reasoning with Latent Structure Refinement for Document-Level Relation Extraction"
127 stars 22 forks source link

关于数据集 #35

Open MingYangi opened 3 years ago

MingYangi commented 3 years ago

请问，CDR数据集去哪里下载

nanguoshun commented 3 years ago

Hi @SeaYM , the CDR dataset is available at https://github.com/JHnlp/BioCreative-V-CDR-Corpus

MingYangi commented 3 years ago

谢谢

MingYangi commented 3 years ago

gen_data.py中第38行有加载‘id2rel.json’，但是DocRE中没有这个文件呀

nanguoshun commented 3 years ago

可以参考DocRED的格式生成这个dictionary文件。因为CDR只有2个relations，所以可以手动创建这个文件。

On Tue, Dec 8, 2020 at 4:22 PM SeaYM notifications@github.com wrote:

gen_data.py中第38行有加载‘id2rel.json’，但是DocRE中没有这个文件呀

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/nanguoshun/LSR/issues/35#issuecomment-740463499, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABBGN7WYB6MC6F5HXZXVLBTSTXO5DANCNFSM4UFUGXLQ .

-- Best Wishes Guoshun Nan

MingYangi commented 3 years ago

其实我一直有点疑惑的是，最后的迭代精化 ，是 执行完 文档编码、结构归纳、推理 这算一次refinement么？然后论文中的两次refinements 是执行了两遍 文档编码、结构归纳、推理么?是这样的意思吗？