fin_sentiment: Money Control Stock News Sentiment Tool

Installing

Install the dependencies by creating the Conda environment fin_sentiment from the given environment.yml file and activating it.

conda env create -f environment.yml
conda activate fin_sentiment

You could face issues with nltk punkt. If that occurs, please do following in python shell:

nltk.download('punkt')

Since we cannot store large pre-trained ML models on GitHub, please download model into models folder.

Getting sentiment

We provide a script to quickly get sentiment predictions.

From the command line, simply run:

python process.py --levels <number of nested levels> --output <file name>

The above script generates a raw json file aggegrating all the links and a summary html file with sentiment of each article.

Please note that the first time you run this script it attempts to download data files and could take a bit longer to run.