Event news detection and citizens community structure for disaster management in social networks
| DOI | https://doi.org/10.1108/OIR-03-2018-0091 |
| Date | 11 February 2019 |
| Published date | 11 February 2019 |
| Pages | 113-132 |
| Author | Radhia Toujani,Jalel Akaichi |
| 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 |
Event news detection and citizens
community structure for disaster
management in social networks
Radhia Toujani
ISG Tunis, University of Tunisia, Tunis, Tunisia, and
Jalel Akaichi
College of Computer Science, University of Bisha, Bisha, Saudi Arabia
Abstract
Purpose –Nowadays, the event detection is so important in gathering news from social media. Indeed, it is
widelyemployed by journaliststo generate early alertsof reported stories.In order to incorporate availabledata
on socialmedia into a news story,journalists mustmanually process, compileand verify the newscontent within
a very shorttime span. Despite itsutility and importance,this process is time-consumingand labor-intensivefor
media organizations. Becauseof the afore-mentionedreason and as social mediaprovides an essentialsource of
data usedas a support for professionaljournalists, the purposeof this paper is to proposethe citizen clustering
techniquewhich allows the community ofjournalists and media professionals to documentnews during crises.
Design/methodology/approach –Theauthors develop, in thisstudy, an approach for naturalhazard events
news detectionand danger citizen’groups clustering based on threemajor steps. In the first stage,the authors
present a pipeline of several natural language processing tasks: event trigger detection, applied to recuperate
potential event triggers; namedentity recognition, used for thedetection and recognition ofevent participants
related to the extracted event triggers; and, ultimately, a dependency analysis between all the extracted data.
Analyzingthe ambiguity and the vaguenessof similarity of newsplays a key role in event detection. This issue
was ignored in traditional event detection techniques. To this end, in the second step of our approach,
the authorsapply fuzzy sets techniqueson these extracted eventsto enhance the clusteringquality and remove
the vaguenessof the extractedinformation. Then,the defined degreeof citizens’danger is injected as input t o the
introduced citizens clustering methodin order to detect citizens’communitieswith close disaster degrees.
Findings –Empirical results indicate that homogeneous and compact citizen’clusters can be detected using
the suggested event detection method. It can also be observed that event news can be analyzed efficiently
using the fuzzy theory. In addition, the proposed visualization process plays a crucial role in data journalism,
as it is used to analyze event news, as well as in the final presentation of detected danger citizens’clusters.
Originality/value –The introduced citizens clustering method is profitable for journalists and editors to
better judge the veracity of social media content, navigate the overwhelming, identify eyewitnesses and
contextualize the event. The empirical analysis results illustrate the efficiency of the developed method for
both real and artificial networks.
Keywords Hierarchical clustering, Risk assessment, Social network analysis, Event detection,
Citizens’community structure, Social news
Paper type Research paper
1. Introduction
The use of social media has dramatically increased in recent years, which produced novel
networked publics for citizen-generated content considered as an essential source of data used
as a supportfor professionaljournalists (Deborahet al., 2017). In fact, the increase of employing
social mediaprofessionals has become important to document news duringcrises, war zones,
politicalelections, sports events, etc. (Brandtzaeg et al., 2016; Stephens-Davidowitz and Pinker,
2017). Most existing works mining social media for event detection and news gathering
addressedthe problem of noiseand burst detection.Indeed, detecting and filtering non-relevant
or noisy content is importantfor isolated users to generate timely and relevant content. Burst
detection issue plays an important role in complex analysis. Moreover, community analysis Online Information Review
Vol. 43 No. 1, 2019
pp. 113-132
© Emerald PublishingLimited
1468-4527
DOI 10.1108/OIR-03-2018-0091
Received 15 March 2018
Revised 4 July 2018
25 September 2018
Accepted 26 September 2018
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/1468-4527.htm
This paper forms part of a special section “Social media mining for journalism”.
113
Disaster
management
in social
networks
technique is generally used to enhance events filtering. Visualization is also essential in data
journalism as the enormity of information needs to be systematically organized into a format
that is easy to communicate. In this context, we develop, in this study, a practical tool to al low
journalists and news readers to recognize newsworthy topics based on message streams
without being plagued by carrying out a dependency analysis between all extracted news. In
fact, most of the proposed research works dealt essentially with the predefinition of a set of
keywords and did not investigate the integrity of these terms, which made difficult, for
journalists, to ensure the accuracy of the reports gathered from social media. Another
complication is that declaring and depictingevent news, whichmay be fuzzy, unclear,incorrect
and incomplete, by citizens’social network can produce various difficulties concerning the
extracted news quality and meaning. To solve the limits associated with verifying event
content and its sources, we integrated the fuzzy theory into a natural language processing
(NLP) method to detect both known and unknown news events, check information and its
sources and contextualize the event. Besides, the majority of the existing techniques focus on
identifying and classifying news events manually, which degrades considerably the
effectiveness of event detection process. To deal with the previously-mentioned issues, ou r
techniquerelies on the burst and noiseevent detection to isolate non-relevant users witha low
degree of dangerand to merge citizens generating timely and relevantcontent. The proposed
clustering event method allows journalists to better judge the event news content. The
performed studies introduced tools having dashboard-style interfaces with complex data
graphics, which is so attractive for some professional users. Thus, visualization process is
suggested, in this paper, to detect event news using of power BI dashboards.
The remainder of this paper is organized as follows: In Section 2, we present the
literature review about using social media to gather news by expert journalists. Then, in
Section 3, we illustrate our employed approach to evaluate both natural dangerous event
news detection and citizens’clustering method. Finally, the experiments are demonstrated
and a brief description of the visualization of the obtained results is provided.
2. Related areas
The importance of the event detection task in gathering news from social media (Dou et al.,
2012) was clearly shown in the literature. Sayyadi et al. (2009) designed a community
model in order to discover events using keywords, noun, phrases and named entities.
Phuvipadawat and Murata (2010) demonstrated the key value of text messages by
counting retweets and detecting popular terms like nouns and verbs. Their study was
further extended by applying a simple tf-idf scheme employed to specify concepts
similarity (Phuvipadawat and Murata, 2011). Afterwards, entities were identified by
utilizing the Stanford Named Entity Recognizer to determine communities and similar
message groups. Twitcident was introduced, in Abel et al. (2012), to broadcast emergency
services, identify incidents, extract a set of related keywords and finally gather related
updates from social media. Ozdikis et al. (2012) detected events employing hashtags
through clustering them and identifying semantic similarities among hashtags. Shi et al.
(2017) depicted a new cosine measure-based event similarity detection technique to
evaluate the correlation between events.
Moreover, generated content displayed on social media is so significant in the procedure
of capturing news events in order to classify and verify stories. In this context, we can
mention the work in Boididou et al. (2018) where authors compared three verification
methods. The first technique focused on employing textual patterns for the extraction of
claims about whether a tweet is fake or real. The second method is based on using
information in which the credibility proves the similarity of event news topic extracted from
tweets. The third technique applied semi-supervised learning scheme that affects the
decisions of two distinct event news credibility classifiers. Experiments performed on
114
OIR
43,1
To continue reading
Request your trial