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

observing real world on twitter
SMART_READER_LITE
LIVE PREVIEW

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


slide-1
SLIDE 1

www.helsinki.fi/yliopisto

Observing real world on Twitter

21.5.2010 1

Ossi Karkulahti Joint work with Jussi Kangasharju and Lasse Nordgren Department of Computer Science University of Helsinki

slide-2
SLIDE 2

www.helsinki.fi/yliopisto

  • Introduction
  • Tale of two cities
  • Topical results
  • Conclusion & future

21.5.2010 2

Outline

slide-3
SLIDE 3

www.helsinki.fi/yliopisto

  • 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.

21.5.2010 3

Introduction

slide-4
SLIDE 4

www.helsinki.fi/yliopisto

  • Motivation

– To understand better the reasons of

users to create tweets, and see if the reasons correspond to real-life situations

– Adaptive content distribution

21.5.2010 4

Introduction

slide-5
SLIDE 5

www.helsinki.fi/yliopisto

  • 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

21.5.2010 5

Introduction

slide-6
SLIDE 6

www.helsinki.fi/yliopisto

  • 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

21.5.2010 6

Tale of two cities

slide-7
SLIDE 7

www.helsinki.fi/yliopisto

  • Results:

– Daily pattern – Hourly pattern – Incidents – Statistics

21.5.2010 7

Tale of two cities

slide-8
SLIDE 8

www.helsinki.fi/yliopisto

Liverpool Madrid

21.5.2010 8

Daily Pattern

Average per day 29307

Mon Tue Wed Thu Fri Sat Sun 5000 10000 15000 20000 25000 30000 35000 40000 Mon Tue Wed Thur Fri Sat Sun 5000 10000 15000 20000 25000 30000 35000 40000

Average per day 34281

slide-9
SLIDE 9

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

Hour Percentage of tweets

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

Hourly Pattern

Liverpool Madrid

slide-10
SLIDE 10

www.helsinki.fi/yliopisto

Liverpool-Reading 1-2, FA Cup, 13th January 2010

21.5.2010 10

Liverpool

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

slide-11
SLIDE 11

www.helsinki.fi/yliopisto

The Brit Awards,February 16th, 8pm ->

21.5.2010 11

Liverpool

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

slide-12
SLIDE 12

www.helsinki.fi/yliopisto

Real Madrid-Barcelona 0-2, La Liga, April 10th

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

slide-13
SLIDE 13

www.helsinki.fi/yliopisto

Madrid: Barcelona:

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

slide-14
SLIDE 14

www.helsinki.fi/yliopisto

  • Liverpool

– On average every sixth tweet has at

least one link

– Every 15th tweet is a retweet

  • Madrid

– On average every third tweet has at

least one link

– Every tenth tweet is a retweet

21.5.2010 14

Links & Retweets

slide-15
SLIDE 15

www.helsinki.fi/yliopisto

  • 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

21.5.2010 15

Users

slide-16
SLIDE 16

www.helsinki.fi/yliopisto

  • 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

21.5.2010 16

Users

slide-17
SLIDE 17

www.helsinki.fi/yliopisto

  • 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

  • lympic games

21.5.2010 17

Topical Results

slide-18
SLIDE 18

www.helsinki.fi/yliopisto 21.5.2010 18

Topical Results

slide-19
SLIDE 19

www.helsinki.fi/yliopisto 21.5.2010 19

Topical Results

slide-20
SLIDE 20

www.helsinki.fi/yliopisto 21.5.2010 20

Topical Results

slide-21
SLIDE 21

www.helsinki.fi/yliopisto

  • 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

21.5.2010 21

Conclusion

slide-22
SLIDE 22

www.helsinki.fi/yliopisto

  • More cities
  • Natural language analysis
  • World Cup 2010
  • Comparison against other social

media services

21.5.2010 22

Future