Balancing acts:

Geopolitical pressures and justifications of the European Commission's digital policies

Code Repository for TRIAS project paper #1

by Christian Rauh & Milan Schröder

WZB Berlin Social Science Center

Scripts:

• pipeline.R: full data analysis pipeline

Text preparation

• 00_connect_DB.R: Establish database connection
• 01_read_corpus.R: Read EC Corpus data from database
• 02_clean_corpus.R: Clean EC Corpus data
• 03_tokenize_corpus.R: Individual tokens of EC corpus (mild pre-processing)
• 04_corpus_descriptives.Rmd: Descriptives on EC corpus

Extract data

• 10_semantic_similarity_weights.R: Extracts different semantic similarity weights on token level from pre-trained GLOVE 6b 300d word embedding model (digitality, economy, security, lib. rights, conflictual/cooperative language)

• 11_semantic_similarity_coding.R: Applies semantic similarity weights to tokenized EC communication texts and aggregates corresponding scores to paragraph level

• 12_12_semantic_similiarity_digital_validation.R: Compare various cut-offs in semantic similarity coding for "digitality" against human codes

• XXX 1x_digitality_zeroshot.ipynb: alternative simple zeroshot classification for benchmarking

• XXX 12_create_validation_sample.R: validation sample creation

• XXX 13_scaling_validation_results.Rmd: validation sample results & cutoff selection

• 20_country_mentions.R: identification of country mentions in EC communication texts along newsmap dictionary

• XXX 21_country_mentions_descriptives.Rmd: descriptives on country identification

Analyse data

• XXX 30_analysis.Rmd: ...

Data:

• ...