Skip to content

Latest commit

 

History

History
14 lines (10 loc) · 423 Bytes

README.md

File metadata and controls

14 lines (10 loc) · 423 Bytes

sentiment_comments_zh

A Chinese comments corpus Sentiment analysis.

inspired by Kaggle Yelp competition, and reversion it to Chinese datasets.

reqirement: keras, tensorflow or theano, scikit-learn ,numpy and so on.

key point: embedding, cnn,binary-class.

to do list:

  1. use pretrained word2vec as inputs.
  2. use pandas to preprocess data faster.
  3. modify model structure.
  4. have a try rnn based sentiment analysis.