A twitter dataset of 2.7B+ tweet IDs across 9+ languages (ongoing collection since Jan 2020), collected by Chen, Lerman, and Ferrara (USC Information Sciences Institute). Each record contains tweet IDs only — full content requires hydration. Largest ongoing COVID-19 Twitter collection with multilingual coverage and continuous updates since pandemic onset. Suitable for pandemic discourse, public health surveillance, infodemic analysis, multilingual NLP.
A Telegram (German-language conspiracy channels) dataset of 63,908,365 messages from 4,404 unique channels, collected by Hagen, Wiencierz, and Freudenthaler via Telegram API. Each record contains message UUID, channel name, author, reply structure, conspiracy label, language, media type (photo/video/voice), audio/video duration, forwarding metadata, duplicate flag. Schwurbelarchiv is a large-scale German Telegram corpus with 3.1% of messages labeled as conspiracy content, 23.8% reply edges enabling thread analysis, and rich media metadata across 4,404 channels. Suitable for conspiracy theory detection, German NLP, social network analysis, Telegram platform research, misinformation propagation, reply network analysis, media sharing patterns.
A Gab (alt-right social network) dataset of 22,112,812 Gab posts from January 2018 (13,871,421 original posts, 8,241,391 reposts), collected by Zannettou et al. via Gab API (2018). Each record contains post body text, username, verified flag, published_at, post type (post/repost), topic category, language, likes/reposts/replies counts, sentiment score, isPopular. 22,112,812 Gab posts from January 2018 — one of the earliest large-scale Gab datasets, capturing the platform during its growth phase as an alt-right Twitter alternative. Suitable for alt-right discourse analysis, fringe platform research, meme propagation, hate speech detection, cross-platform information flow, deplatforming research.
The wikilink network of the German Wikipedia, as distributed by KONECT (wikipedia_link_de), built from the Wikimedia dumps: 3,225,565 articles and 81,626,917 directed links between them, where a link from A to B means article A links to article B (article namespace only; 31,722 pages link to themselves). Counts are those of ECHO's copy of the KONECT archive; KONECT's current release lists 3,603,726 articles and 96,865,851 links, including articles without links. One of eight language editions on ECHO (English, German, French, Italian, Japanese, Polish, Portuguese, Russian). Useful for knowledge-graph and hyperlink-structure analysis, link prediction, community detection and comparison across language editions.
A Bluesky Social dataset of approximately 1.8 million public posts, collected by Alpin Dale from the AT Protocol firehose API in 2024. Each record contains post text, author handle, timestamp, image presence flag, and reply-to URI for constructing directed reply-thread networks. multilingual corpus from a decentralized AT Protocol platform with English dominant alongside Spanish, Portuguese, Japanese, and other languages. Suitable for decentralized social media analysis, reply network structure, and multilingual content modeling.
A Twitter dataset of 2,280 user accounts, 368,028 tweets, collected by Yang and Menczer (OSoMe/Indiana University), 2023. Each record contains user_id, label (bot/human), dataset source, full tweet objects with text and metadata. First large-scale dataset of AI-powered social bots (Fox8 botnet) with GPT-generated tweet content; includes human baseline accounts for comparison.. Suitable for AI bot detection, LLM-generated content identification, social botnet analysis..
A Bluesky (2024 presidential debate coverage) dataset of 279,249 bluesky posts from June 25 - July 2, 2024, collected by Zenodo contributors via public API. Each record contains post text, user ID, post ID, creation date, hashtags, mentioned users, URLs, post type (post/reply/repost), linked post, root post, language. 279,249 bluesky posts covering the 2024 Biden-Trump presidential debate with pre-computed sentence embeddings. Suitable for political communication, debate analysis, cross-platform comparison, decentralized social media research, NLP with embeddings.
A Mastodon (2024 presidential debate coverage) dataset of 73,789 mastodon posts from June 25 - July 2, 2024, collected by Zenodo contributors via public API. Each record contains post text, user ID, post ID, creation date, hashtags, mentioned users, URLs, post type (post/reply/repost), linked post, root post, language. 73,789 mastodon posts covering the 2024 Biden-Trump presidential debate with pre-computed sentence embeddings. Suitable for political communication, debate analysis, cross-platform comparison, decentralized social media research, NLP with embeddings.
A TextDetox 2024/2025 shared task (15 languages) dataset of 71,374 posts across 15 languages with binary toxicity labels, collected by Dementieva et al. for PAN/CLEF TextDetox 2024. Each record contains post text, binary toxicity label (0=non-toxic, 1=toxic), language. Combined multilingual toxicity benchmark covering 15 languages including low-resource languages (Amharic, Tatar, Hindglish) — 71,374 balanced samples total. Suitable for multilingual toxicity detection, cross-lingual transfer learning, low-resource NLP, content moderation benchmarking.
A Twitch.tv (live streaming platform) dataset of 34,118 users and 214,556 undirected friendship edges across 6 language communities (DE, ENGB, ES, FR, PTBR, RU) merged into one cumulative graph, collected by Rozemberczki et al. via Twitch API, distributed via MUSAE/SNAP. Each record contains undirected friendship edges between Twitch streamers within each language community; node IDs namespaced by language to avoid collision across subgraphs. The dataset is notable for its multilingual streaming platform social network covering 6 distinct language communities, enabling cross-community structural comparison and multilingual graph embedding research. Intended for research into community detection, graph embedding, node classification, cross-platform social network analysis, and multilingual network comparison.
A YouTube (with cross-platform spread on Twitter and Facebook) dataset of 7,659 removed YouTube videos with multilingual titles and descriptions, 249,346 associated tweets, and 18,897,277 Facebook shares, collected by Knuutila (2021) via YouTube Data API and CrowdTangle, tracking videos shared on public social media that were subsequently removed by YouTube for COVID-19 misinformation. Each record contains video titles and descriptions in multiple languages, view counts and subscriber counts where recoverable, Facebook reactions/comments/shares, Twitter and Facebook post ID lists for social spread tracking, removal and publication timestamps, and Wayback Machine archive URLs. The dataset is notable for its one of the few public datasets documenting the complete lifecycle of platform-removed health misinformation, linking video metadata to cross-platform social spread and enabling research into both content and dissemination patterns of COVID-19 misinformation. Intended for research into health misinformation detection, content moderation research, cross-platform information spread analysis, COVID-19 infodemic studies, multilingual misinformation narrative analysis, and platform policy impact assessment.
A TextDetox 2024 shared task (German) dataset of 5,000 German posts (2,500 toxic, 2,500 non-toxic), collected by Dementieva et al. for PAN/CLEF TextDetox 2024. Each record contains post text, binary toxicity label (0=non-toxic, 1=toxic). TextDetox German provides 5,000 balanced binary toxicity annotations — part of a 15-language multilingual benchmark for cross-lingual toxicity detection research. Suitable for toxicity detection, multilingual content moderation, cross-lingual NLP, hate speech classification, text detoxification.
A CLEF CheckThat! 2025 Task 2 (german) dataset of 386 social media post and normalized claim pairs, collected by CLEF CheckThat! Lab organizers (2025). Each record contains social media post text containing a factual claim, normalized canonical claim statement. CLEF 2025 CheckThat! Task 2 (german) provides 386 pairs of social media posts and their normalized factual claims, enabling claim normalization and canonicalization research. Suitable for claim normalization, fact-checking automation, NLP text generation, multilingual claim extraction, social media fact verification.
Twitter (X)
Telegram
YouTube
Gab
Wikipedia
Bluesky
Mastodon
Twitch
Facebook