This is a social-network-mining project to quantify public opinion on certain company. The whole idea of this project is based on the class CIS600 of Syracuse University.
- time
- pandas
- wordcloud
- nltk
- matplotlib
- re
These can be installed via: pip install twitter pip install times pip install pandas pip install wordcloud pip install nltk pip install matplotlib pip install re
You must first set up a Twitter APP Input this data into the config file with your keys.
For using the preprocessing module, please copy the QPO_preprocessing.py file to the path where you run the training program and import QPO_preprocessing
To apply the preprocessing, you need to create a preprocessing model first. For example: model = pp_model('Dell.csv') The only parameter is the file path which you want to train.
After creation, you can apply the basic process model.processing(), the sampling based on user model.userBasedSample(df, 1) and the sampling based on date model.dateBasedSample(df). Or, you can just use model.full_processing() to get all the preprocessing effects.
For emotion recognition: h5py==2.9.0 Keras==1.1.0 numpy==1.16.0 pandas==0.24.1 python-dateutil==2.8.0 pytz==2018.9 PyYAML==5.1 scipy==1.2.1 six==1.12.0 Theano==1.0.4
For another method please download the newest textblob. Use any python IDE and modify the path for tweets input in file SA.py and demo.py