Priceless Tweets!

16
Priceless Tweets! A study on twitter messages A study on twitter messages posted during crisis: Black d Satur day. Suku Sinnappan Cathy Farrel Elizabeth Stewart Swinburne University of Technology Swinburne University of Technology # Locusts # Locusts # Fire # Fire # Flood # Flood # Fire # Fire

description

Presented at ACIS Dec 2010

Transcript of Priceless Tweets!

Page 1: Priceless Tweets!

Priceless Tweets! A study on twitter messagesA study on twitter messages posted during crisis: Black 

dSaturday.Suku SinnappanCathy Farrely

Elizabeth StewartSwinburne University of TechnologySwinburne University of Technology

# Locusts# Locusts

# Fire# Fire

# Flood# Flood

# Fire# Fire

Page 2: Priceless Tweets!

OutlineOutline

• Background

• ObjectivesObjectives

• Motivation

• Significance

• Literature and MethodologyLiterature and Methodology

• Analysis and Discussion

• Future research

Page 3: Priceless Tweets!

BackgroundBackground

• Web 2.0 as a channel for communication

• 2009 ‐ 3 billion tweets worldwide2009  3 billion tweets worldwide

• 9000 tweets a minute (O’Brien 2009) 

• Lightweight (140 char), crowdsourcing 

• Twitter has been used for numerous purposesTwitter has been used for numerous purposes– Politics, PR, marketing, education, Personal, 

t l ti hi ffi i l i i tcustomer relationship, official, crisis etc

Page 4: Priceless Tweets!

ObjectivesObjectives

• Web 2.0 has not been used to effectively communicate during crisis despite its potential g p pin Australia

• Can Twitter help and How?• Can Twitter help and How?

• What kind of information is tweeted during crisis in Australia?

• Can Twitter be used an alternative channel for• Can Twitter be used an alternative channel for communication during crisis?

Page 5: Priceless Tweets!

MotivationMotivation

• There are few motivations leading up to the studyy– Web 2.0 used for official info dissemination but not communication despite huge public uptakenot communication despite huge public uptake

– The failure of traditional communication channels during crisisduring crisis

– Inadequate ground level information during crisis

– Valuable information untapped from Web 2.0

Page 6: Priceless Tweets!

SignificanceSignificance

• An alternative communicate channel (both ways) for large scale disastery ) g

• Reporting valuable information effectively (public and official communication)(public and official communication)

• Potential reduction in casualties or damage

Page 7: Priceless Tweets!

LiteratureLiterature

T itt b t t diti l di t d li t d t• Twitter beats traditional media to deliver up to date information

• Iran election, Mumbai Terrorist, Bangkok Red Shirt protest, , , g p ,Gaza Flotilla. Including yesterday’s student hostage crisis at Marinette, Wisconsin

• Sichuan earthquake (Gabarain 2008) Californian fires• Sichuan earthquake (Gabarain, 2008), Californian fires (Wagner, 2008, Sutton Et al, 2008), New England Ice Storm (Ragan, 2009), Gulf of Mexico Hurricane (Janega, 2008), C l N i i M (P l ti 2008) d Mi i i iCyclone Nagris in Myanmar (Palatino, 2008) and Mississippi Hurricane (Brown, 2009), Red River Flood (Starbird et al. 2010), Mass emergency (Hughes et al. 2009), Haiti (Palen et al. 2010), Seattle Violent Crisis (Herevin and Zach, 2010).

Page 8: Priceless Tweets!

MethodologyMethodology

• Gather tweets during crisis to analyse

• 2009 Victoria bush fires using ‘fires’ as used by2009 Victoria bush fires using  fires  as used by Herevin and Zach (2010)

N d d f i k f• Needed to form categories to make sense of the tweets – used Namaan et al (2010) as base

• 1684 tweets between 6nd – 14th Feb 2009• 1684 tweets between 6 – 14 Feb, 2009

• 705 unique users

Page 9: Priceless Tweets!

Analysis and Discussion (1/7)Analysis and Discussion (1/7)No. of

• Top 10 (as shown)

• Only 2 officialNo. User Name No. of

Tweets Followers (as

of 19 July 2010)Only 2 official 

channels

E l f ffi i l

)1 774melbourne 63 13,845 2 geehall1 49 655 3 retrogrrl 29 1,223

• Example of official dissemination

retrogrrl 29 ,4 SassyCupcakes 20 148 5 tellyworth 20 91 6 duncanriley 19 8,525

• Not reported in royal commission

duncanriley 19 ,7 problogger 19 100,859 8 stephenedgar 18 1,331 9 wolfcat 18 2,585royal commission 9 wolfcat 18 2,585

10 abcnewsMelbrne 16 3,996

Page 10: Priceless Tweets!

Analysis and Discussion (2/7)Analysis and Discussion (2/7)

• Validating data collected against another crisis (Uganda pub bombing after world cup 2010)( g p g p )

• Tweets were categorised by batches of 100s based on Namaan et al while referring tobased on Namaan et al while referring to Sutton et al. 2008, Hughes et al. 2009, Heverin and Zach 2010Heverin and Zach, 2010.

Page 11: Priceless Tweets!

Analysis and Discussion (3/7)

No. Categories of tweets No. of tweets

1 W b Li k b i h 400

• * Naaman et al. (# 6‐13)

• Opinions/Complaints and Statements/Random 

1 Web Links to maps, websites, other resources, 400

2 Geographical data i.e. location, name of a place 730

Other factual description i.e. Time, Figure, Cardinal directions, Other organisations

Thoughts were merged as they were inapplicable

• Each tweet was allowed to 3 (formal names), types of crisis (flood, fire, tsunami etc), communication outlets

(radio, TV, telephone), climate related information875

4 Suggestions, request followers to act, actionable information, and current status update 264

be coded into more than 1 category due to information

• Ave 1.98 categories, 4  update

5 Seeking information, help or answers 119

6* Statements or random thoughts – a comment by the user on a situation 331

7* M t t t b t th i ‘ti d d t’ 218

g ,iterations 

• 2 dropped (#12 ,#13), new(#1‐#5) 7* Me now – current status about the user i.e. ‘tired and upset’ 218

8* Presence maintenance – instant display of user existence i.e. ‘I’m backkkk’ 3

9* Other’s status – informing followers about other’s situation 2

( )

• Tweets that were less informative (# 7,8,11) were miniscule as compared to

10* Communication with Followers – reply tweets, forwarding tweets, linking others in a message 491

11* Self promotion i.e. ‘Check out my blog’ 1

12* Anecdote me – an exciting-funny incident about the user 0

miniscule as compared to 80% reported by Namaan et al  ‐ while 65% were informative 12 Anecdote me an exciting-funny incident about the user 0

13* Anecdote others - an exciting-funny incident about others 0• Tweets need to be decoded in context

Page 12: Priceless Tweets!

Analysis and Discussion (4/7)Analysis and Discussion (4/7)• Objective 2 could Twitter be used during times of crisis as an alternative• Objective 2 – could Twitter be used during times of crisis as an alternative 

communication channel?• 12% carried links ‐ Google maps, Country Fire Authority updates, 

Weatherzone com Bureau of Meteorology ABC net auWeatherzone.com, Bureau of Meteorology, ABC.net.au, Incidentalert.com.au, TheAge.com.au and social sites such as Flickr and Photobucket.

• 22% had Geo‐location data  ‐ “Ah, too late. Yackandanda is under alert, , ,and the Beechworth fires are heading south‐east. 100+ homes gone, 14+ dead. #bushfires”. 

• 25% had rich ground level information–– “fires in narre north, two streets away from my house. shit.”– “ADSL has been down all afternoon, Sky was surreal bronze colour, due to 

fires”“t t t f Y l d d th th t l d t d th fi ”– “two routes out of Yea are closed and the other two lead towards other fires”

– “Fighting fires !”

Page 13: Priceless Tweets!

Analysis and Discussion (5/7)Analysis and Discussion (5/7)

O ti li i th t t i ‘#’ t i h• Operationalising the tweets using ‘#’ to increase search ability. Eg Palen et al (2010) Haiti project (EPIC)‐– #name (name),#name (name), – #loc (location) , – #num (amount or capacity), – #contact (email, phone, link, other) or #con,– #photo (link to photo), #source (source of info) or #src #status (e g open closed– #source (source of info) or #src, #status (e.g., open, closed, injured, etc.), 

– #date (date or time), – #info (other information).

• Contextualise the # terms to suit disaster and region

Page 14: Priceless Tweets!

Analysis and Discussion (6/7)Analysis and Discussion (6/7)

• F• For eg –– “Anyone have news on whether the fires are near Healesville?”– “#help Anyone have news on whether the #crisis fires are near #loc 

H l ill ?”Healesville?”.• 5% had direct actionable data ‐ “@royrussel ABC Gippsland 

Transmitter burned by fires tune into 531AM , 828AM, 104.7FM, 105 5FM 90 7FM f l l d t @774M lb ”105.5FM or 90.7FM for local update ‐@774Melbourne”,

• 119 tweets were seeking for information/ assistance (similar to phone calls to SES/CFA/Police)– “I'm looking after an evacu‐cat from Warburton to help out 

@firstdogonmoon 's friend. Bush fires affect pets too!” – “Anyone have news on whether the fires are near Healesville?”

– ” Does anyone know if the fires touched Smith's Gully? Worried about an ex‐colleague.”

Page 15: Priceless Tweets!

Analysis and Discussion (7/7)Analysis and Discussion (7/7)

– Tweets also prove that Twitter was used for crisis communication during the bushfires by – public and state authorities :

• “CFA urges those not directly affected by the Vic fires to get info from ABC Melbourne http://bit.ly/MsGj (expand) rather than the CFA site”

f d ill b d l b d– A way forward will be to deploy bots to detect searchable # tags, aggregate information with geo‐l i d h l i f b i b 2 0location tag and channel info by using web 2.0 itself . This offers an organic solution to users.

Page 16: Priceless Tweets!

Future ResearchFuture Research

Th d h k• There are many ways to extend the current work• Exploring ways to aggregate information (both direction)

Ed i b h bli d l l h i i b W b 2 0• Educating both public and  local authorities about Web 2.0 communication (device and software)

• Involving other web 2 0 tools – Facebook Myspace OrkutInvolving other web 2.0 tools  Facebook, Myspace, Orkutetc

• Applying the same study to flood, hurricane, terrorist, earth quakes, or other natural disasters such locust attacks. 

• Validating the quality of information i e spam and info• Validating the quality of information i.e. spam and info accuracy (eg. Flood info ABC invalid link)