For the seminar "Solving the Climate Crisis with Text Mining" at the Hasso Plattner Institute Potsdam we developed this project with the goal of identifying the policy positions of German political parties on the topic of coal power. We are using a dataset of speeches from the German Bundestag going back to 1949, all of them containing some mention of a word with the substring "Kohle".
The project is separated into two main approaches, Wordfish and
Wordscores. For both approaches we provide an example
usage file in the src directory. The approach implementation as well as utility functions can be found in the
respective src/wordfish or src/wordscores directories. General utility functions are to be found in the
src/utils directory.
We also looked into running traditional, dictionary-based sentiment analysis on the data, but found it to not be
applicable to this problem. The code for this attempt can be found in src/sentiment_analysis.
The sandbox.R file was only used for development and testing, so it is not up-to-date and should thus not be used as
an entry point to our project code.