Text Sentiment in the Age of Enlightenment

This repository contains the code for the analysis conducted in the paper Text Sentiment in the Age of Enlightenment

The Data

Spectator Journals

We work with TEI encoded XML Files available at: https://gams.uni-graz.at/mws. All XML files need to be placed in the same dictionary. Further you have to correctly set paths to files in the respective Jupyter Notebooks.

Sentiment Dictionaries

The used dictionaries are available at: https://sites.google.com/site/datascienceslab/projects/multilingualsentiment. These dictionaries were created and analyzed by Chen and Skiena [1].

The Code

Code is available as Jupyter Notebooks. Execute the notebooks in the following order:

01_extract_data.ipynb

Used to extract texts and additional meta data from the XML files. You have to configure two paths:

1. data_dir: Pointing to the directory containing the XML files.
2. results_dir: Pointing to the directory where results should be stored.

02_conduct_analysis.ipynb

Used to conduct the analysis. Make sure that you set the dataframe_input_path to the pickled dataframe file from the extraction and to point sentiment_dir to the directory containing the downloaded sentiment dictionaries.

Supplementary Material

You can find the top 100 central words according to degree, betweenness and closeness centrality for German, French, Italian and Spanish networks in the topwords directory.

References

[1] Chen, Y., Skiena, S.: Building sentiment lexicons for all major languages. In: Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers). pp. 383–389 (2014)