Sentiment analysis - Githubissues

elitcloud / elit

🔠 Evolution of Language and Information Technology

https://elit.cloud

Other

47 stars 7 forks source link

Sentiment analysis #15

Closed bgshin closed 6 years ago

bgshin commented 6 years ago

PIP packages

Keras, tensorflow(-gpu)

Required files

Six files are required to be located in the subdirectory of elit/resources/
- trained word2vec
  - elit/resources/sentiment/w2v/w2v-400-amazon.gnsm
  - elit/resources/sentiment/w2v/w2v-400-amazon.gnsm.syn0.npy
  - elit/resources/sentiment/w2v/w2v-400-semevaltrndev.gnsm
  - elit/resources/sentiment/w2v/w2v-400-semevaltrndev.gnsm.syn0.npy
- trained twitter model:
  - elit/resources/sentiment/model/s17-400-v2
  - elit/resources/sentiment/model/sst3-400-v2
- All files are at ainos02:/tmp/sentiment/*

Options for users to choose CPU or GPU

In setup.py, we need to add a parameter to tell if an user wants to run it on CPU or GPU.
if it is CPU, then one of the dependencies should be "tensorflow", or if it is GPU, then it should be "tensorflow-gpu"

Usage of the SentimentAnalysis module

# initialize the class for SST (movie review)
sa = SSTModel()
# decode
y, att, raw_att = sa.decode(tokenized_sentences)