Dataset for the "The Schwurbelarchiv: a German Language Telegram dataset for the Study of Conspiracy Theories" paper. Metadata.

Dataset for the "The Schwurbelarchiv: a German Language Telegram dataset for the Study of Conspiracy Theories" paper. Metadata. The data is freely available. The file schwurbelarchiv_final_fwd_authors_metadata.parquet (6.9 GB) has 63,908,365 rows and contains all columns of the full dataset except the four text fields. It is sufficient to reproduce all results reported in the paper. The fields are: uuid: unique identifier for each row (string)folder_id: internal folder identifier for the group/channel (string)chat_name: name of the group/channel (string)chat_type: type of chat ("channel" or "group") based on the number of different authors within a group (string)author: hashed name of the author (string)link_url: link as found in the HTML a-tag (string)media_file: name of the attached media file (string)media_file_type: "voice message", "video", or "photo" (string)message_id: ID of the message in the respective group (string)posting_date: date and time of the message in the group (UTC-aware timestamp)replied_to: message_id of the message that has been replied to (string)website: website domain, if link_url refers to a website (string)fwd_link_url: link from the forwarded message (string)fwd_media_file: internal path of the forwarded media file (string)fwd_media_file_type: "voice message", "video", or "photo" (string)fwd_posting_date: date and time of the message in the original group (UTC-aware timestamp)fwd_author: hashed name of the original author of the forwarded message (string)message_lang: language detected in the message (string)is_conspiracy: whether the message was classified as conspiracy content (boolean)audio_length_s: duration of an audio file in seconds (float)audio_size_mb: file size of the attached audio file in MB (float)audio_lang: language detected in the audio transcription (string)video_length_s: duration of a video file in seconds (float)video_lang: language detected in the video transcription (string)is_duplicate: boolean flag marking rows that duplicate another row due to redundant scraping of that group; such rows are kept and flagged instead of deleted because the replied_to column may reference only one of them (boolean) It particularly does not contain the fields: message: content of the messagefwd_message: content of the forwarded messageaudio_transcription: transcribed content of speech-containing audio files, covering both original and forwarded voice messagesvideo_transcription: transcribed content of speech-containing video files, covering both original and forwarded videos