TWIMPACT 2011 Real-Time Twitter Analysis TWIMPACT analysis ~ 2 ms data delivery ~ 1000 ms 1.3b tweets 35.8m users 11 GB / day 1 day 1 s 1 minute 1 hour 50 3000 180,000 5M 3.3TB 0.6 TB © 2011 TWIMPACT UG (haftungsbeschränkt) in cooperation with Berlin University of Technology 880,000 page links 2,500 400 14,500 102,000 each day 5,400,000 retweets 140,000 media links TWIMPACT analyses the data by 1. parsing each JSON message 2. extracting meta-data for users, tweets, location ... 3. analyzing the tweet text language 4. matching a retweet in previous tweets 5. extracting mentions users, hashtags and links 6. updating retweeted and mentioned user relations 7. updating trend counts for all artifacts 8. updating indices (full-text, users, all artifacts) 9. removing irrelevant data from memory retweet activity Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0 12k 24k 36k 48k 60k iPad 2 Steve Jobs iPad 2 Steve Jobs resigned Steve Jobs died 0 600 1200 1800 2400 3000 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Google Plus launched retweets / week 58 24 25 24 26 38 37 31 33 29 29 26 25 32 35 39 37 46 55 49 45 43 36 35 32 32 29 31 30 28 27 26 24 94 106 117 42 4 3 4 5 3 3 4 4 31 37 42 31 33 35 32 28 34 27 30 27 27 52 47 39 50 45 44 37 38 38 30 23 28 31 31 36 36 35 36 30 33 31 35 37 4 3 4 5 10 29 109 84 71 55 57 51 16 41 52 51 52 49 41 37 30 30 45 39 33 45 40 43 35 41 42 45 49 45 50 47 50 56 55 53 53 55 51 44 41 44 39 35 44 65 57 53 48 54 47 42 49 47 40 49 30 79 113 128 119 90 90 86 60 72 77 68 59 59 61 59 58 49 46 57 62 68 70 64 68 55 61 56 61 56 50 48 88 82 78 75 78 91 93 95 99 82 86 83 67 72 67 66 55 57 49 42 40 109 72 66 57 71 68 58 53 52 57 48 41 12 35 60 63 75 78 93 88 89 87 81 81 91 134 144 158 164 146 147 138 121 138 124 127 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec damnitstrue thenoteboook justinbieber ugglytruth xstrology ihatequotes metrotv meatschool radiogalaufm ohteenquotes teenlifefeel lilduval sextrology lord_voldemort7 teenlovefeel epictweets_ zodiacfacts viatumblr retweetifs revrunwisdom top 20 impact most retweeted Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec @wendys RT for a good cause. Each Retweet sends 50¢ to help kids in foster care. #TreatItFwd 3k 6k 9k 12k 15k 3k 6k 9k 12k 15k 3k 6k 9k 12k 15k @mentionto #MentionTo your friend who's never be on time and lazy to do anything. @justinbieber life has it's ups and downs but u guys r always there 4 me. i will make my mistakes but i promise 2 continue 2 grow with u and try 2 do rt additional graphs and indices identify relations between users, retweets, mentions, languages and many other artifacts 500,000 user relations 15,000,000 user/rt relations 1,000 historic snapshots every 8 hours Mikio L. Braun, Matthias L. Jugel, Klaus-Robert Müller http://twimpact.com http://beta.twimpact.com/ely/