Mining German politicians' tweets

tl;dr

This is an ongoing project on the analyses of some tweets of some German politicians.

In this repository, you will find code and output. Note that NO data is made available, as there are legal concerns in sharing Tweets. However, in acccordance with the regulations of Twitter, I have made the "dehydrated" tweets available (IDs only; the IDs can be re-hydrated using the open Twitter API, or a number of tools).

Analyses are conducted using R.

Install and run:

1. Install R: https://stackoverflow.com/questions/20457290/installing-r-with-homebrew
2. Download RStudio: https://www.rstudio.com/
3. git clone https://github.com/sebastiansauer/polits_tweet_mining.git
4. Open polit_tweets_mining.Rproj
5. Install Packages: Within RStudio, go to Tools>Install Packages…, type in the packages needed from the files to be found within /code (e.g. tidyverse)
6. The order of the analyses is noted chronologically in the file names (ie., start with file preceded by 01_.

Collecting profiles/twitter accounts

Twitter accounts were collected based on these lists:

https://twitter.com/dw_politics/lists/german-politicians/members https://www.afd.de/partei/bundesvorstand/

Collecting tweets

Tweets were collected via the Twitter API, mediated via the R package TwitteR. For each entry in the list of accounts, the maximum number of tweets allowed (n=3200) were collected. However, in some situations, the API delivers fewer tweets.

In sum, ~320k tweets were collected.

Oldest tweet: March 2013 Most recent tweet: Aug 2017

Please note that the tweets are stored in dehydrated form due to the legal requirements of Twitter. Tweets can be re-hydrated using the Twitter API (or via the respective R package).

Analyses

Metadata of tweets and text of tweets were processed (after initial makeup/preparation). Stopwords and sentiment analyses (in prep) are included.

Some result presentation

Check out this slides for some result presentation.