LEMON stands for LExicon-MemOry-augmented-Ner. (https://arxiv.org/abs/1912.08282)

• Python: 3.6 or higher
• PyTorch: 1.0

The dataset can be configured in the dataset.py.

A mixed version of the CoNLL format, with each character and its label(segmentation, POS-tag and NER-tag) for one line. Sentences are split with a null line.

中 B B-ni B-ORG
共 M M-ni M-ORG
中 M M-ni M-ORG
央 E E-ni E-ORG
致 S S-v O
中 B B-ns B-ORG
国 E E-ns M-ORG
致 B B-ni M-ORG
公 M M-ni M-ORG
党 E E-ni M-ORG
十 B B-j M-ORG
一 M M-j M-ORG
大 E E-j E-ORG
的 S S-u O
贺 B B-n O
词 E E-n O

##Pretrained Embeddings

• Character embeddings: gigaword_chn.all.a2b.uni.ite50.vec
• Lexicon embeddings: ctb.50d.vec

##Running

python main.py