Information retrieval of a disaster event from cross-platform social media

DOIhttps://doi.org/10.1108/IDD-01-2017-0003
Pages220-226
Date20 November 2017
Published date20 November 2017
AuthorShi Shen,Nikita Murzintcev,Changqing Song,Changxiu Cheng
Subject MatterLibrary & information science,Library & information services,Lending,Document delivery,Collection building & management,Stock revision,Consortia
Information retrieval of a disaster event from
cross-platform social media
Shi Shen
State Key Laboratory of Earth Surface Processes and Resource Ecology, Beijing Normal University, Beijing, China and
Key Laboratory of Environmental Change and Natural Disaster, Beijing Normal University, Beijing, China
and Center for Geodata and Analysis, Faculty of Geographical Science, Beijing Normal University, Beijing, China
Nikita Murzintcev
LREIS, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing, China, and
Changqing Song and Changxiu Cheng
State Key Laboratory of Earth Surface Processes and Resource Ecology, Beijing Normal University, Beijing, China and
Key Laboratory of Environmental Change and Natural Disaster, Beijing Normal University, Beijing, China
and Center for Geodata and Analysis, Faculty of Geographical Science, Beijing Normal University, Beijing, China
Abstract
Purpose – The purpose of this study is to propose a method to retrieve data on an event based on a preliminary collection of event-specific hashtags.
Design/methodology/approach – Extra knowledge, or a list of events with recorded features that can be used to characterize an event and
separate it from other simultaneously occurring social phenomena, is employed. The first step involves the estimation and use of the impact area
to retrieve messages from Twitter. This is followed by an extraction of hashtags from these messages. After that, the noisy hashtags would be filtered
out by some heuristic rules. Finally, hashtags are used to collect relevant messages from not only Twitter but also other social media platforms.
Findings – The proposed method has high selectivity and is able to collect distinct sets of hashtags even for similar simultaneous events. In addition,
spatial and temporal features are sufficient to improve collecting information of disaster events.
Originality/value – This work discusses a method of information retrieval of an event from cross-platform social media. The proposed method can
be applied to other studies of geographically related events.
Keywords Information retrieval, SVM, Typhoon Haiyan, Cross-media, Event-related, Hashtags
Paper type Research paper
1. Introduction
Social media is a computer-mediated tool that allows people
and companies to exchange news, information, ideas, pictures
and videos in virtual communities. Social media provides
information on a wide range of social activity and contains
inter-agent connections that are characterized by fast response
from a local to a user event. Therefore, social media is often
used to analyze human behavior in dangerous situations. For
example, Lu and Brelsford (2014) use a complex network to
investigate the digital human behavior change in the 2011
Japanese Earthquake and Tsunami, and Olteanu et al. (2015)
found that emergency managers and official emergency
response agencies are increasingly using social media as part of
their information gathering processes. In such kind of studies,
the first task is to find proper information in social big data.
Historically, most of the studies on information extraction
of disaster events were conducted on Twitter, as it provides
open (but limited) access to its data (Regalado et al., 2015;
Grasso and Crisci, 2016;Lachlan et al., 2015;Salfinger et al.,
2016). The common practice is to use keywords to retrieve
messages and then classify them to filter the noisy records out
to increase the relevance of data.
The correct choice of the keywords is a crucial part, as social
media is a big data source and limits the amount of records to be
downloaded and processed in the following procedure. The
keywords are usually chosen manually. Indeed, Olteanu et al.
(2014) proposed a method to build event-specific lexicons to
simplify data-retrieval process. First, awareness systems used
manual classification, and then modern systems combined
crowdsourcing and machine-learning methods for classification
(Ashktorab et al., 2014;Chowdhury et al., 2013;Imran et al.,
2014).
A similar approach can be used to retrieve data in other
social media platforms (Facebook, Instagram, Pinterest, etc.),
but it should be adapted for every case individually
(Benkhelifa and Laallam, 2016;Kagaya and Aizawa, 2015;
Yang et al., 2015). These platforms use different programming
interfaces and data formats and contain different types of
The current issue and full text archive of this journal is available on
Emerald Insight at: www.emeraldinsight.com/2398-6247.htm
Information Discovery and Delivery
45/4 (2017) 220–226
© Emerald Publishing Limited [ISSN 2398-6247]
[DOI 10.1108/IDD-01-2017-0003]
This work was supported by National Natural Science Foundation of
China [grant numbers 41771537] and the Fundamental Research Funds
for the Central Universities.
Received 7 January 2017
Revised 22 September 2017
Accepted 22 September 2017
220

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT