nlp-server
Usage
Run fasttext server:
nohup python -u serve_fasttext.py --model /path/to/wiki.en.bin --port 8980 &Query word vectors (python):
import requests, json
url = 'http://127.0.0.1:8980/fasttext'
tokens = ['apple', 'asdf@andrew.cmu.edu']
resp = requests.get(url, params={'tokens': json.dumps(tokens)}).json()
vectors = resp['vectors']
assert vectors.shape == (2, 300)Query word vectors (shell):
curl http://127.0.0.1:8980/fasttext?tokens=[%22apple%22]Run SRL server:
nohup python -u serve_srl.py --port 8981 &Get SRL predictions:
import requests
url = 'http://127.0.0.1:8981/srl'
requests.post(url, json=[{'sentence': 'I love you'}, {'sentence': 'asdf'}]).json()
# [{'verbs': [{'verb': 'love',
# 'description': '[ARG0: I] [V: love] [ARG1: you]',
# 'tags': ['B-ARG0', 'B-V', 'B-ARG1']}],
# 'words': ['I', 'love', 'you']},
# {'verbs': [], 'words': ['asdf']}]Run dependency parser:
nohup python -u serve_parser.py --port 8982 &Get dependency parse:
import requests
url = 'http://127.0.0.1:8982/parse'
requests.post(url, json=[{'sentence': 'I love you'}, {'sentence': 'asdf'}]).json()
# [{'words': ['I', 'love', 'you'],
# 'pos': ['PRON', 'VERB', 'PRON'],
# 'predicted_dependencies': ['nsubj', 'root', 'dep'],
# ...Get verb sense SRL predictions:
import requests
url = 'http://127.0.0.1:8983/cogcomp_verb_srl'
requests.post(url, json=[{'sentence': 'I love you'}]).json()
# [{'verbs': [{'verb': 'love',
# 'sense': '1.0',
# 'description': '[ARG0: I] [V: love] [ARG1: you]',
# 'tags': ['B-ARG0', 'B-V', 'B-ARG1']}],
# 'words': ['I', 'love', 'you']}]Get nominal sense SRL predictions:
import requests
url = 'http://127.0.0.1:8984/cogcomp_nom_srl'
requests.post(url, json=[{'sentence': 'Twitter confirms sale of company to Elon Musk for $44 billion.'}]).json()
# [{'nominals': [{'nominal': 'sale',
# 'sense': '01',
# 'predicate_index': [2],
# 'description': '[ARG0: Twitter] [Support: confirms] sale [ARG1: of company] [ARG2: to Elon Musk] [ARG3: for $ 44 billion]',
# 'tags': ['B-ARG0','B-Support','O','B-ARG1','I-ARG1','B-ARG2','I-ARG2','I-ARG2','B-ARG3','I-ARG3','I-ARG3','I-ARG3']}],
# 'words': ['Twitter','confirms','sale','of','company','to','Elon','Musk','for','$','44','billion']}]