Leveraging cross-media analytics to detect events and mine opinions for emergency management
| Date | 14 August 2017 |
| Published date | 14 August 2017 |
| DOI | https://doi.org/10.1108/OIR-08-2015-0286 |
| Pages | 487-506 |
| Author | Wei Xu,Lingyu Liu,Wei Shang |
| Subject Matter | Library & information science,Information behaviour & retrieval,Collection building & management,Bibliometrics,Databases,Information & knowledge management,Information & communications technology,Internet,Records management & preservation,Document management |
Leveraging cross-media
analytics to detect events
and mine opinions for
emergency management
Wei Xu and Lingyu Liu
Renmin University of China, Beijing, China, and
Wei Shang
Academy of Mathematics and Systems Science, Chinese Academy of Sciences,
Beijing, China
Abstract
Purpose –Timely detection of emergency events and effective tracking of corresponding public opinions are
critical in emergency management. As media are immediate sources of information on emergencies,
the purpose of this paper is to propose cross-media analytics to detect and track emergency events and
provide decision support for government and emergency management departments.
Design/methodology/approach –In this paper, a novel emergency event detection and opinion mining
methodis proposed for emergency managementusing cross-mediaanalytics.In the proposed approach, an event
detection module is constructed to discover emergency events based on cross-media analytics, and after the
detected event is confirmed as an emergency event, an opinion mining module is used to analyze public
sentimentsand then generatepublic sentimenttime series forearly warning via a semanticexpansiontechnique.
Findings –Empirical results indicate that a specific emergency can be detected and that public opinion can
be tracked effectively and efficiently using cross-media analytics. In addition, the proposed system can be
used for decision support and real-time response for government and emergency management departments.
Research limitations/implications –This paper takes full advantage of cross-media information and
proposes novel emergency event detection and opinion mining methods for emergency management using
cross-media analytics. The empirical analysis results illustrate the efficiency of the proposed method.
Practical implications –The proposed method can be applied for detection of emergency events and
tracking of public opinions for emergency decision support and governmental real-time response.
Originality/value –This research work contributes to the design of a decision support system for
emergency event detection and opinion mining. In the proposed approaches, emergency events are detected
by leveraging cross-media analytics, and public sentiments are measured using an auto-expansion of the
domain dictionary in the field of emergency management to eliminate the misclassification of the general
dictionary and to make the quantization more accurate.
Keywords Opinion mining, Cross-media analytics, Emergence management, Event detection,
Semantic expansion
Paper type Research paper
Introduction
Emergencies with destructive effects may threaten people’s daily lives , as well as the
peace of social order, if they cannot be detected in a timely manner and effectively
controlled. As widely studied in previous research, emergencies can be classified as
natural disasters, accidental disasters, public health incidents and social security
incidents, which have also been defined as situations that pose risks to society stability
(Green and Kolesar, 2004).
Online Information Review
Vol. 41 No. 4, 2017
pp. 487-506
© Emerald PublishingLimited
1468-4527
DOI 10.1108/OIR-08-2015-0286
Received 30 August 2015
Revised 29 January 2016
Accepted 26 February 2016
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/1468-4527.htm
This work was supported by National Natural Science Foundation of China (Grant Nos 71301163,
71571180), Humanities and Social Sciences Foundation of the Ministry of Education (Nos 14YJA630075,
15YJA630068), Hebei Social Science Fund (HB13GL021), the Fundamental Research Funds for the Central
Universities, and the Research Funds of Renmin University of China (Nos 10XNK159, 15XNLQ08).
487
Emergency
management
Therefore, emergency management is crucial for government and other decision makers
to perform to avoid risks. According to the existing literature, emergency management is
generally applied by governments and organizations to avoid further disasters and reduce
the negative impact on the public based on methods of supervision, collaboration and
leadership (Waugh and Streib, 2006).
Characterized by relevance, complexity, infrequency and unpredictability, public
emergencies often cause a serious threat to social stability and national security.
Unconventional emergencies have four time periods, the latent period, burst period,
sustaining period and disappeared period, and each stage has different characteristics
( Jiang et al., 2010). With guidance regarding the four periods’evolution, emergency events
can be detected, and corresponding public opinion can be tracked in real time within the first
three periods. In this manner, measures can be taken to control an emergency or prevent it
from entering into the next period more easily.
Meanwhile, in the era of Web 2.0, with the increasing growth of digital content on the
internet, user-generated content (UGC, e.g., blogs, Twitter feeds, web forums and
micro-blogs) is universally generated and spread, expressing and exchanging people’s
opinions and sentiments. Most UGC contains rich sentiment and obvious topics that are
useful for analyzing people’s behaviors and public opinions, which are obviously relevant in
the field of security. For example, previous research (Fang and Peress, 2009) confirmed that
stocks with no media coverage could yield higher future returns than ones with heavy
media coverage. Other research (Tetlock, 2007) has shown that investors will be influenced
by the internet media’s emotion, which will lead to the decrease of stock prices.
Therefore, more and more researchers now consider combining traditional emergency
management and online information modeling. Previous research (Fogli and Guida, 2013)
proposed a decision support system designed based on a knowledge center for emergency
management. Others have contributed to rumor theoretic analysis based on online
communities during social crises (Oh et al., 2013; Petz et al., 2013). Moreover, a prediction
module based on content analysis of interestingness was proposed to predict the retweet
probability of bad news (Naveed et al., 2011). In the domain of hotspot detection, an online
forum hotspot detection module based on K-means clustering and support vector machines
has been proposed (Li and Wu, 2010), and also an ontology-based event detection is
developed (SanMiguel et al., 2009). Furthermore, an ontological subscription and blocking
system is applied to detect information system post-development change requests conflicts
and alleviate information overload in social blogs.
Therefore, in our research, natural language processing techniques for analysis of online
information,such as keyword extraction and sentimentanalysis, are applied to UGC research
to identify and classify its topics and emotions.Sentiment analysis is always used in judging
whether the content is objective or subjective and whetherthe content is positive or negative.
Among the techniques of sentiment analysis, two main streams are proposed.The first one is
lexicon-based techniques, and the other ismachine-learning techniques. Almost all sentiment
analysisapproaches are based on the bag-of-words assumption(Blei et al., 2003), which means
that the order of words in a document can be neglected in somenatural language processing.
As general sentiment-classification techniques, lexicon-based techniques emphasize more
about the words’or phrases’meanings and theirsentiment combination in the entire content
based on a specific lexicon. In the direction of lexicon-based approaches, the way to build an
emotional dictionary is being explored by more and more researchers.
In previous studies of lexicon-based techniques for sentiment analysis, a researcher
(Turney,2001) measured words’and phrases’sentiment orientation byapplying the result of a
subtraction between the value of the mutual information of the phrase and the word “excellent”
(which has what is called positive polarity) as well as the mutual information of the phrase and
the word “poor”(which has what is called negative polarity). Then, a semi-supervised way to
488
OIR
41,4
To continue reading
Request your trial