Conversation

As promised, here is a tweet id dump for a full day of Twitter activity. This constitutes approximately > 99.95% of all available tweets and retweets for the sample day. There are 431,022,103 tweet ids in this file. files.pushshift.io/tweet_ids_samp The format of the data is...

Quote Tweet

JSON Baumgartner

@jasonbaumgartne

Nov 25, 2020

Would any researchers be interested in getting a full day's worth of tweet ids that they can rehydrate to get a data on what a day in the life of Twitter looks like? In order to abide by Twitter's TOS, this would have to be for research purposes. I would only be able to provide

Show this thread

4:48 PM · Nov 26, 2020Twitter Web App

Replying to

tweet id, lang, is_retweet, is_reply, is_verified, num_hashtags, num_mentions, num_urls, tweet text length. After reviewing the Twitter TOS for research purposes, I believe this dump abides by the new Twitter rules for disseminating tweet ids for research purposes.

JSON Baumgartner

@jasonbaumgartne

Nov 26, 2020

This data covers exactly a 24 hour period (in UTC time). The average volume of tweets is 4,989 tweets per second during the timeframe provided.

Replying to

This is great, Jason! Thanks for all the nice resources you produce.

Replying to

Thanks for sharing the data. How did you collect the tweet ids? by reconstructing the firehose method?

Replying to

Awesome, thanks for doing this Jason.

Replying to

That might be coolest thing in 2020! It can provide answers to many questions I have! Thank you very much :)

Replying to

Super interesting data to explore, thank you! Looking at the tweets over time, I noticed these spikes, shortly after a full hour, the tallest at 15 hours (UTC). What do you think that is? Automated tweets at the full hour I guess, but what's going on at 15h? Noon in Brazil?

Replying to

and

Yes probably automation. Those spikes even occur each 30 mins. I would love to throw my text stream clustering algorithm on the data.