没有合适的资源?快使用搜索试试~ 我知道了~
资源推荐
资源详情
资源评论
Association for Information Systems
AIS Electronic Library (AISeL)
PACIS 2011 Proceedings
Pacific Asia Conference on Information Systems
(PACIS)
9 July 2011
Social Media Data Mining: A Social Network
Analysis Of Tweets During The 2010-2011
Australian Floods
France Cheong
RMIT University, france.cheong@rmit.edu.au
Christopher Cheong
RMIT University, christopher.cheong@rmit.edu.au
ISBN: [978-1-86435-644-1]; Full paper
This material is brought to you by the Pacific Asia Conference on Information Systems (PACIS) at AIS Electronic Library (AISeL). It has been
accepted for inclusion in PACIS 2011 Proceedings by an authorized administrator of AIS Electronic Library (AISeL). For more information, please
contact elibrary@aisnet.org.
Recommended Citation
Cheong, France and Cheong, Christopher, "Social Media Data Mining: A Social Network Analysis Of Tweets During The 2010-2011
Australian Floods" (2011). PACIS 2011 Proceedings. Paper 46.
http://aisel.aisnet.org/pacis2011/46
SOCIAL MEDIA DATA MINING: A SOCIAL NETWORK
ANALYSIS OF TWEETS DURING THE AUSTRALIAN 2010-
2011 FLOODS
France Cheong, School of Business IT and Logistics, RMIT University, Melbourne 3000,
Victoria, Australia, france.cheong@rmit.edu.au
Christopher Cheong, School of Business IT and Logistics, RMIT University, Melbourne
3000, Victoria, Australia, christopher.cheong@rmit.edu.au
Abstract
Using tweets extracted from Twitter during the Australian 2010-2011 floods, social network analysis
techniques were used to generate and analyse the online networks that emerged at that time. The aim
was to develop an understanding of the online communities for the Queensland, New South Wales and
Victorian floods in order to identify active players and their effectiveness in disseminating critical
information. A secondary goal was to identify important online resources disseminated by these
communities. Important and effective players during the Queensland floods were found to be: local
authorities (mainly the Queensland Police Services), political personalities (Queensland Premier,
Prime Minister, Opposition Leader, Member of Parliament), social media volunteers, traditional
media reporters, and people from not-for-profit, humanitarian, and community associations. A range
of important resources were identified during the Queensland flood; however, they appeared to be of
a more general information nature rather than vital information and updates on the disaster. Unlike
Queensland, there was no evidence of Twitter activity from the part of local authorities and the
government in the New South Wales and Victorian floods. Furthermore, the level of Twitter activity
during the NSW floods was almost nil. Most of the active players during the NSW and Victorian
floods were volunteers who were active during the Queensland floods. Given the positive results
obtained by the active involvement of the local authorities and government officials in Queensland,
and the increasing adoption of Twitter in other parts of the world for emergency situations, it seems
reasonable to push for greater adoption of Twitter from local and federal authorities Australia-wide
during periods of mass emergencies.
Keywords: social network analysis, text mining, social media, mass emergencies.
1
Cheong and Cheong: Social Media Data Mining: A Social Network Analysis Of Tweets Dur
Published by AIS Electronic Library (AISeL), 2011
1 INTRODUCTION
Australia experienced its worst flooding disasters in 2010 and 2011 with a series of floods occurring
in several states between March 2010 and February 2011. These floods are considered the worst
flood events in the history of the various states affected. First, there were the Queensland floods of
March 2010 causing major flooding in south western and central Queensland (The Courier-Mail,
2010), followed by the Victorian floods of September 2010 damaging major regional towns such as
Ballarat and Benalla (ABC News, 2010). Next, there was the flooding of the River Gascoyne in
Western Australia in December 2010 damaging homes in Carnavon (Australian Bureau of
Meteorology, 2010), the December 2010-January 2011 floods of Queensland causing three-quarters of
the state to be declared a disaster zone (Brisbane Times, 2011), and ending with the New South Wales
of January 2011 (Nine News) and the Victorian floods of February 2011 (The Age, 2011).
In times of mass emergencies, a phenomenon known as collective behaviour becomes apparent
(Dynes & Quarantelli, 1968). It consists of socio-behaviours that include intensified information
search and information contagion (Starbird, Palen, Hughes, & Vieweg, 2010). In these situations,
people want to know where exactly their families and friends are as not being able to reach them or
knowing they might not be able to contact you can be very frightening moments during these
situations. Information is critical during emergencies as the availability of immediate information can
save lives. People share information about approaching threats, where to evacuate, where to go for
help, etc. Not only do they want to know about the destruction that has occurred, but they are also
eager to help those affected by giving a helping hand and raise funds from donations. Thus, there is a
need to keep abreast of the latest developments, however, this is difficult since information produced
under crisis situations is usually scattered and of varying quality.
Social media is media used for social interaction. They are enabled by communication technologies
such as the web and smartphones and they turn communication into an interactive dialogue
(Wikipedia, 2011). Interactions on social media being highly distributed, decentralised and occurring
in real time, they provide the necessary breadth and immediacy of information required in times of
emergencies (Palen & Vieweg, 2008). Since social media offer a uniquely rapid and powerful way to
disseminate information, accurate and inaccurate, good and bad spread equally alike as incorrect
information can spread like wild fire. However, there is indication that social networks tend to favour
valid information over rumours (Castillo, Mendoza, & Poblete, 2011).
Twitter and Facebook are good examples of social media useful in crisis situations since they provide
vital information as they are happening. Twitter is a micro-blogging service, a form of lightweight
chat allowing users to post and exchange short 140-character-long messages known as tweets.
Although most tweets are conversation and chatter, they are also used to share relevant information
and report news (Castillo, et al., 2011). Twitter is becoming a valuable tool in disaster and emergency
situations as there is increasing evidence that it is not just a social network, it is also a news service
(Yates & Paquette, 2011). In emergency situations, tweets provide either first-person observations or
bring relevant knowledge from external sources (Vieweg, 2010). Information from official and
reputable sources is regarded as valuable and hence is actively sought and propagated. Other users
then elaborate and synthesize this pool of information to produce derived interpretations.
During the Mumbai terror attacks of 2008, online users voluntarily created a Twitter page
(http://www.twitter.com/Mumbai) to update and share situational information on the attacks (Oh,
Agrawal, & Rao, 2010). A study found that 52.6% of tweeted H1NI-related material in 2009 to be
related to news and information on swine flu (Chew, Eysenbach, & Sampson, 2010). Twitter was used
to provide time-critical information about tsunami alerts, missing and deceased people, availability of
services, road conditions among other topics related to the catastrophe hours after the Chile
earthquake of Santiago in 2010 (Mendoza, Poblete, & Castillo, 2010). During the Haiti earthquake,
Twitter was used to create awareness about the disaster and mobilize people to help (Yates &
Paquette, 2011).
2
PACIS 2011 Proceedings, Art. 46 [2011]
http://aisel.aisnet.org/pacis2011/46
Social Network Analysis (SNA) is a sociological approach for analysing patterns of relationships and
interactions between social actors in order to discover underlying social structure such as: central
nodes that act as hubs, leaders or gatekeepers; highly connected groups; and patterns of interactions
between groups (Wasserman & Faust, 1994). SNA has been used to study social interaction in a wide
range of domains. Examples include: collaboration networks (Newman, 2001), directors of companies
(Davis & Greve, 1997; Davis, Yoo, & Baker, 2003), organisational behaviour (Borgatti & Foster,
2003), inter-organisational relations (Stuart, 1998), computer-mediated communications (Garton,
Haythornthwaite, & Wellman, 1999), and many others.
In this study, we propose to use SNA to study the community of Twitter users disseminating
information during the crisis caused by the Australian floods in 2010-2011 in order to reveal
interesting patterns and features within this online community. With the help of SNA, we hope to
develop an understanding of the online community that was active during that period by answering
the following questions: What was the online social behaviour during the flood period? In particular,
who were the active players in communicating information and how effective were they? What type
of information was of importance? How can the information discovered be useful for the management
of such situations in the future?
2 RELATED WORK
Given the usefulness of Twitter as a communication and interaction platform for disaster and
emergency situations, it is not surprising to find out that a number of studies have been performed to
analyse tweets posted under such conditions.
A content analysis of 1684 tweets collected during Black Saturday, Australia’s worst fire disaster
found that these tweets contained actionable factual information contrasting with claims that the
contents of tweets are of no value as they are mostly chatter (Sinnappan, Farrell, & Stewart, 2010).
In the case of the Mumbai terrorist attack, Oh et al. (2010) performed a qualitative analysis to argue
that the terrorists monitored tweets posted by networked citizens to their advantage as they utilised
situational information to mount attacks against civilians.
Over 2 million tweets were collected during a period of about 8 months during the 2009 H1N1
outbreak and manual content analysis was performed on a random sample of 5,395 tweets (Chew, et
al., 2010). Content analysis revealed that 52.6% of the posts were resources-related and the most
popular resources were news websites followed by web pages of government and health agencies.
The credibility of information contained in tweets was the subject of research during the Chile 2010
earthquake, manual (Mendoza, et al., 2010) and automatic credibility analysis (Castillo, et al., 2011)
were performed. Manual examination of the veracity of information disseminated on a small number
of tweets during that critical event showed that false rumours tend to be questioned much more than
confirmed truths. An automatic classifier was built using features extracted from “trending topics”
and was found to be able to classify tweets as credible or not credible with precision and recall in the
range of 70% to 80%.
Tweets posted during the 2009 Red River floods were analysed to identify the mechanisms for
information production, distribution and organization (Starbird, et al., 2010). It was found that the
production of new information on Twitter is by means of derivative activities such as directing,
relaying, synthesizing, and redistribution and is additionally complemented by socio-technical
innovation. Twitter activity during the Red River floods and Oklahoma grassfires were analysed to
identify information that may contribute to situational awareness (Vieweg, 2010). The objective of the
study was to identify the features of the information generated for the development of framework to
inform the design and implementation of software systems employing information extraction
strategies. A prescriptive tweet-based syntax was proposed to increase the utility of information
generated during emergencies (Starbird & Stamberger, 2010).
3
Cheong and Cheong: Social Media Data Mining: A Social Network Analysis Of Tweets Dur
Published by AIS Electronic Library (AISeL), 2011
剩余16页未读,继续阅读
资源评论
go2girl
- 粉丝: 0
- 资源: 1
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功