A fast and efficient chinese FAQ approach with One shot learning (Siamese) and TF-IDF techniques.
- tensorflow 1.14.0
- keras 2.2.4
See more details in faq_siamese.py
if __name__ == '__main__':
# train model
train()
# show candidate
print('*' * 30)
exps = ['为什么植物会指南','小孩为什么不能吃补品中药？']
for exp in exps:
for x in ShowCandidate(exp):
print(x)
print('*'*30)
# answer a question
print(Answer('为何植物会吃石头'))