Observing real world on Twitter Ossi Karkulahti Joint work with - PowerPoint PPT Presentation

Observing real world on Twitter Ossi Karkulahti Joint work with Jussi Kangasharju and Lasse Nordgren Department of Computer Science University of Helsinki www.helsinki.fi/yliopisto 21.5.2010 1

Outline ● Introduction ● Tale of two cities ● Topical results ● Conclusion & future www.helsinki.fi/yliopisto 21.5.2010 2

Introduction ● Twitter Twitter is a real-time information network powered by – people all around the world that lets you share and discover what’s happening now. Twitter asks “what’s happening” and makes the answer spread across the globe to millions, immediately. ● Tweet A tweet is a post or status update on Twitter. The – maximum size of a tweet is 140 characters. www.helsinki.fi/yliopisto 21.5.2010 3

Introduction ● Motivation – To understand better the reasons of users to create tweets, and see if the reasons correspond to real-life situations – Adaptive content distribution www.helsinki.fi/yliopisto 21.5.2010 4

Introduction ● We have collected more than 5 million different tweets, by using two different methods: – Topical keywords , such as “H1N1” and “Olympics” – Based on the location of the users, e.g. Liverpool, Madrid, Rome etc. ● Indicated in the profile or by geotag www.helsinki.fi/yliopisto 21.5.2010 5

Tale of two cities ● During Jan - April 2010 we have gathered tweets from two cities: – Liverpool, UK (~3.2 million tweets) – Madrid, the capital of Spain (~3.4) ● Location-based method www.helsinki.fi/yliopisto 21.5.2010 6

Tale of two cities ● Results: – Daily pattern – Hourly pattern – Incidents – Statistics www.helsinki.fi/yliopisto 21.5.2010 7

Daily Pattern Liverpool Madrid 40000 40000 35000 35000 30000 30000 25000 25000 20000 20000 15000 15000 10000 10000 5000 5000 0 0 Mon Tue Wed Thu Fri Sat Sun Mon Tue Wed Thur Fri Sat Sun Average per day 34281 Average per day 29307 www.helsinki.fi/yliopisto 21.5.2010 8

Hourly Pattern 0,09 0,08 Liverpool 0,07 0,06 Percentage of tweets 0,05 0,04 0,03 0,02 0,01 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Hour 0,09 0,08 Madrid 0,07 0,06 0,05 0,04 0,03 0,02 0,01 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Liverpool Liverpool-Reading 1-2, FA Cup, 13th January 2010 1945 GMT: kick-off 2031: GOAL 1-0 2033: Half-time 1-0 ~2050: 2nd half kick-off 2139: Penalty to Reading 2140: GOAL 1-1 2142: End of 90 mins 1-1 2147: Extra-time begins 2156: GOAL 1-2 2203: Half-time in extra-time 1-2 2221: Full-time in extra-time 1-2 2240: Interviews www.helsinki.fi/yliopisto 21.5.2010 10

Liverpool The Brit Awards,February 16 th , 8pm -> Most common words during 20-23 pm: Rank Count 2 brits 919 17 gaga 336 20 peter 286 21 robbie 285 36 cheryl 216 37 brit 213 45 lady 186 46 awards 183 56 liam 160 61 award 147 69 florence 139 78 music 123 89 williams 112 106 cole 98 108 gallagher 94 120 alicia 88 129 kasabian 84 137 ladygaga 79 152 dizzee 70 www.helsinki.fi/yliopisto 21.5.2010 11

Madrid Real Madrid-Barcelona 0-2, La Liga, April 10th 2000 GMT: kick-off 2032: GOAL 0-1 2047: Half-time 0-1 ~2102: 2nd half kick-off 2112: GOAL 0-2 2150: Full-time 0-2 www.helsinki.fi/yliopisto 21.5.2010 12

Madrid: 2000 GMT: kick-off 2032: GOAL 0-1 2047: Half-time 0-1 ~2102: 2nd half kick-off 2112: GOAL 0-2 2150: Full-time 0-2 Barcelona: www.helsinki.fi/yliopisto

Links & Retweets ● Liverpool – On average every sixth tweet has at least one link – Every 15 th tweet is a retweet ● Madrid – On average every third tweet has at least one link – Every tenth tweet is a retweet www.helsinki.fi/yliopisto 21.5.2010 14

Users ● Liverpool, in total 87761 users – Users with 1000+ tweets: 802 = 1 % ● But 50 % of all tweets – Users with 51-1000 tweets: 6222 = 7 % ● 26 % of all tweets – Users with 2-50 tweets: 36105 = 41 % ● 8 % of all tweets – Users with only one tweet: 44632 = 51 % ● 16 % of all tweets www.helsinki.fi/yliopisto 21.5.2010 15

Users ● Madrid, in total 103 632 users – Users with 1000+ tweets: 813 = 0.8 % ● But 48 % of all tweets – Users with 51-1000 tweets: 7573 = 7 % ● 28 % of all tweets – Users with 2-50 tweets: 44713 = 43 % ● 9 % of all tweets – Users with only one tweet: 50533 = 49 % ● 16 % of all tweets www.helsinki.fi/yliopisto 21.5.2010 16

Topical Results ● We have collected tweets related to the swine flu outburst and the 2010 Winter Olympics with such keywords as: – H1N1, swineflu, and swine flu – Vancouver, olympic, olympics, and olympic games www.helsinki.fi/yliopisto 21.5.2010 17

Topical Results www.helsinki.fi/yliopisto 21.5.2010 18

Topical Results www.helsinki.fi/yliopisto 21.5.2010 19

Topical Results www.helsinki.fi/yliopisto 21.5.2010 20

Conclusion ● The results indicate that – There are regional and cultural differences – The users are tweeting about current events, such as sporting events, awards shows, and topical situations – The user are willing to express both their positive and negative thoughts www.helsinki.fi/yliopisto 21.5.2010 21

Future ● More cities ● Natural language analysis ● World Cup 2010 ● Comparison against other social media services www.helsinki.fi/yliopisto 21.5.2010 22