paper:
Neural Architectures for Named Entity Recognition
End-toEnd Sequence labeling via BLSTM-CNN-CRF
code:
https://github.com/ZhixiuYe/NER-pytorch
python3.6
pytorch
train data at ./dataset/aminer_train.dat
<word> <label>
train model:
python train.py
query:
-
run the server
python server.py
-
GET Method:
http://166.111.5.228:5011/query/<query>
-
Return in json, Example:
http://166.111.5.228:5011/query/search some selection Thomas Edison State College 1902 Goel Shom 's papers {"LOC": ["selection", "1902"], "PER": ["Goel"], "CON": ["Shom's"], "DATE": [], "ORG": ["Edison", "State"], "KEY": ["Thomas", "College"], "O": ["search", "some", "papers"]}
|- train.py
|- debug.py
|- [dir] dataset (word library)
|- [dir] evaluation (help tools when training)
|- [dir] models (well-trained models)