/repol

predicting social media reaction from a news

Primary LanguagePython

Repol

  • I trained a multi-output linear regression model that predicts Ethiopian Facebook users reaction to a political news

  • I scraped facebook using fbcrawl (https://github.com/rugantio/fbcrawl)

  • I scraped data FBC(Fana broadcasting conrporate) news Page

  • I used googletrans a free unofficial google trainslate library to convert from Amahric to English

Limitations

  • Very small Dataset
  • Future Me: THIS REPO HAS A LOT OF PROBLEMS
  • not normalized data
  • bad word representation
  • poor translation

Requirements

python 3.x

Installation

git clone https://github.com/robelkebede/repol.git

pip install -r requirements.txt

Usage

cd ./repol
python repol.py --text="Green Legacy campaign"

Alt text

Training Loss

Alt text

Contributing

Pull request are welcome. specially new data crawlerd from diffrient Facebook news pages and please open an issue first

License

[MIT] (https://choosealicense.com/licenses/mit/)