This repository contains the code for reproducing the results of the analysis of "Whose job is it anyway? An analysis of the sentiment, semantics and contributors to the Danish news coverage of a parental leave reform". The data from the articles is not available for sharing due to restrictions on data sharing from Infomedia. With access to Infomedia, this repository aims to illuminate the specific preprocessing and analysis decisions made, and to make it reproducible for interested parties.

A requirements.txt file is included in the folder named requirements and can be used to rebuild the virtual environment used for the analysis in Python.

1. Follow the instructions and code in data_cleaning/Articles_to_csv.Rmd. This converts the text in the PDFs of articles and converts them to a .csv.
2. Run sentiment_analysis.py to use the BERT_Tone model to classify sentences as either negative, neutral or positive.
3. Run clean_data.py to process and clean the tokens for articles