CooRTweet 2.2.0

• detect_groups() now accepts time_window_low to select a single interval of absolute time differences. Both limits are included; the default lower limit of 0 retains the previous behavior. Invalid or reversed limits now produce an error.
• Replaced deprecated igraph calls in network generation and the paper reproduction script with their current equivalents.
• prep_data() now removes incomplete and duplicate shares by default. Set drop_na = FALSE and/or drop_duplicates = FALSE to retain them. The input data.table is no longer modified by reference.
• The installed package now includes german_elections_sample (97,413 rows; under 1 MB) for multi-platform and multi-modal examples. The full german_elections data is no longer installed; the unchanged data file remains in the source repository under data-raw/, with the paper reproduction script in vignettes/reproduce_examples.Rmd. Scripts that loaded german_elections from the installed package should switch to german_elections_sample or load the full data from the source repository, depending on their purpose.
• Updated the introductory vignette to use the sample, the new defaults, and the correct GitHub installation address.
• generate_coordinated_network() now defaults to subgraph = 1, returning only edges above the edge_weight percentile threshold. Set subgraph = 0 to retain the previous default behavior and return the full graph with threshold indicators. With fast_net = TRUE, the default filters the full network by its edge weights; use subgraph = 2 for the filtered fast network.

CooRTweet 2.1.0

• Added a new multi-modal and multi-platform dataset for testing and demonstration purposes (see documentation for “german_elections” data).
• Updated the vignette
• New function to simulate data

CooRTweet 2.0.2

Edit the function documentation to address specific issues and enhance clarity. This includes removing obsolete information about min_repetition in the detect_groups documentation and correcting typos in the vignettes.

CooRTweet 2.0.1

Fixed a bug that under certain circumstances prevented the number of shares from being calculated correctly with the accounts_stats function.

CooRTweet 2.0.0

A major release with many important breaking changes. Package users upgrading to version 2.0.0 should pay attention to the functions detect_groups and generate_coordinated_network

New features

• completely reworked the code that generates the coordination network. There is now a new function generate_coordinated_network
• The results from detect_groups can now be subset into “faster” subnetworks with the new function flag_speed_share
• The prep_data function helps to bring other social media data into the right shape

Changes

Most changes are related to two areas of improvement: the way how the coordination network is generated and naming conventions used in this package

• detect_coordinated_groups function was renamed to detect_groups, alognside we also renamed the min_repetition to min_participation to better reflect its meaning. The function generally performs still performs the same operation as before.
• generate_network was removed and replaced with generate_coordination_network, which is a completely rewritten approach to the network generation.
• We globally changed the wording of “user” to “account”, which is more faithful to the theoretical framework of this package. The term “account” describes the data much better.
• Therefore, the column formally named id_user (found in many functions) was now renamed to account_id

Dropped Features

We decided to drop the feature to detect cotweeting behaviour based on text similarity for this release. In practical applications, the detect_similar_text function was not performing up to our standards and we are currently looking into a fresh approach to implement such a feature.

CooRTweet 1.4.0

This is a minor release adding new features

New features

• Added detect_similar_text() function that finds tweets based on text similarity (cotweets). Refer to the documentation for details.
• The reshape_tweets() function was extended with the intent = "cotweet", which reformats your data to fit in the detect_similar_text() function