=Paper=
{{Paper
|id=Vol-2293/jist2018pd_paper5
|storemode=property
|title=Content-based Visualization System For Sentiment Analysis On Social Networks
|pdfUrl=https://ceur-ws.org/Vol-2293/jist2018pd_paper5.pdf
|volume=Vol-2293
|authors=Julio Vizcarra,Kouji Kozaki,Miguel Torres Ruiz,Rolando Quintero
|dblpUrl=https://dblp.org/rec/conf/jist/VizcarraKRQ18a
}}
==Content-based Visualization System For Sentiment Analysis On Social Networks==
https://ceur-ws.org/Vol-2293/jist2018pd_paper5.pdf
Content-Based Visualization System For
Sentiment Analysis On Social Networks
1
Julio Vizcarra , Kouji Kozaki
1,∗ 2
,Miguel Torres Ruiz ,Rolando Quintero
2
1
The Institute of Scienti c and Industrial Research (ISIR) Osaka University
Mihogaoka 8-1, Ibaraki, Osaka 567-0047. Japan.
2
Centro de Investigación en Computación CIC , Instituto Politécnico Nacional,
UPALM-Zacatenco, CIC. Building, 07738, Mexico City, Mexico.
Abstract. A content-based visualization system is presented for the
sentiment analysis on social networks. The methodology implemented
was focused on the semantic processing taking into account the content
in the public user's opinions. In our approach the comments were handled
as excerpts of knowledge. During the visualization the social graph is
displayed presenting the polarity and sentiment status for each comment.
Moreover a web mapping tool retrieves comments in a radius based on
the location source(geographic) or concepts related to geographic entities
and spatial relations in the comment(conceptual).
Keywords: sentiment analysis, knowledge engineering, conceptual similarity
1 Introduction
Nowadays the huge information transmitted on social networks has become a
rich source of information for the human understanding as well as a way of
expression where the users share their sentiment status and personal opinions
through comments. The sentiment identi cation can classify comments as posi-
tive or negative(polarity) and unveil emotions such as anger, trust, sadness ,etc.,
on certain topics or users. Moreover the sentiments presented in the opinions can
be relevant in the design of custom services, social plans for public health, mar-
keting, e-commerce,etc. On the basis of these motivations, we developed a web
system for real-time monitoring of sentiment information in social networks for
speci c targets(public events or users). The system is able to display the social
graph structure, sentiment information related as well as retrieve comments by
source's location or words geo-referenced in the text by means of a web mapping.
2 Methodology
The methodology implemented handles the comments as excerpt of the knowl-
edge, in this gap we prioritized the semantic level, sense and meaning of the whole
* Corresponding author.
E-mail address: kozaki@ei.sanken.osaka-u.ac.jp.
2 Julio Vizcarra et al.
comment. The proposal computed semantic similarity measures, conceptual ex-
pansion, graph theory algorithms and disambiguation using a multi domain
knowledge base. The methodology is composed for the following stages: the stage
of social networks discovery retrieves opinions from events or public pro les by
reading their comments. Consequently the stage of knowledge processing con-
structs the formal representation for each comment by using an knowledge base
and graph algorithms such as steiner tree [6] and shortest paths[2]. This mod-
ule carries out processes of automatic knowledge graph construction enhanced
by disambiguation. For processing general knowledge on speci c languages we
considered the English, Japanese and Multilingual Wordnets[1][4][5]. In addition
the dictionaries that provided sentiment information were SentiWordnet [3] and
NRC emotion lexicon [7]. On the other hand the online services for geographic
information processing were WikiData and GeoNames. Finally the stage of sen-
timent analysis estimates the total polarity and main sentiment presented in the
comments. For each concept the polarity is obtained from the knowledge base by
average and the closets sentiment by knowledge graph expansion and shortest
path. The highest polarity and main sentiment are established to a comment.
3 Visualization
This section describes the system that implemented our methodology. We present
examples of the sentiment analysis and their visualization.
In order to explain the sentiment analysis an example was processed from
Twitter in the CNN News account. The comment considered is: "a number of
people feared dead after a dam bursts in kenya with hundreds left homeless
o cials say". The table 1 presents the closest sentiment and a polarity value
assigned by our methodology to each concept.
Id Wordnet-Concept Sentiment with polarity
WN:107449542-n (" are",burst) Sentiment:NRC_fear_NRC_anger|:Polarity:-0.25 ,
WN:107964900-n (homeless) Sentiment:NRC_anticipation_disgust_anger|Polarity:-0.125 ,
WN:107534492-n (fear) Sentiment:NRC_fear,sadness,anger,surprise|Polarity:-0.875 ,
WN:114509110-n (say) NRC_surprise_anticipation|Polarity:0.5
Table 1. Sentiment-Polarity assigned to concepts
Finally the methodology estimates the total polarity and main sentiment
presented in the comment. The values established were for polarity: -0.1875 and
main sentiment: NRC_Anger.
Additionally some relevant results from Twitter account CNN News are pre-
sented. The table 2 de nes the main sentiment and polarity value assigned by
our methodology to the comments. We noticed a better and more trustworthy
classi cation using the basic sentiments instead of polarity (average).
Title Suppressed Due to Excessive Length 3
Sentiment Polarity Comment
trust 0.2916667 This couple found a buried safe containing $52,000 worth of money,
gold and jewelry in their backyard, but didn't keep it
trust -0.15 In an eort to keep conversations and search results on topic, Twitter
announced it will use new "behavioral signals" to push down more
tweets that "distort and detract"
anger 0.04166667 A massive poaching ring in Oregon and Washington is accused of
killing more than 200 animals including deer, bears, cougars, bobcats
and a squirrel
anger 0.041666687 An estimated 239,000 girls under the age of ve die in India each year
due to neglect linked to gender discrimination, a new study nds
sadness 0.25 @CNN Her father had a heart surgery and cant walk so
sadness -0.25 Teen develops 'wet lung' after vaping for just 3 weeks
joy 0.125 I am proud to be a woman and a feminist. The politics of Meghan
Markle
Table 2. Other examples processed in twitter
In the visualization the results are displayed in the system by means social
graphs and web mapping. Regarding the social graph it describes the network's
structure and its sentiment information related to comments by colors. For in-
stance the gure 1 presents the polarity and sentiment graphs for the CNN news
account. Regarding the nodes the darkest blue represents the user target and
light blue for users farther. Particularly in the polarity graph the nodes with
gray color represents neutral comments and the scale between green and red for
positive to negative polarities respectively. On the other hand in the sentiment
graph each comment has a sentiment represented by a dierent color.
Fig. 1. polarity and sentiment graphs
In addition the web mapping tool retrieves comments by location which can
be geographic(location source) or conceptual(Geo-referenced concepts and re-
lations), The gure 2 retrieves comments by conceptual processing using the
keyword Arkansas in a distance of 1000 km. The comment that contains in its
description the concept memphis is retrieved.
4 Julio Vizcarra et al.
Fig. 2. web mapping
4 Conclusions
In this poster a content-based methodology and its implementation were pro-
posed for the sentiment identi cation. The novelty of the presented our approach
is the capability of handling the comments as excerpts of knowledge. We pro-
vided a mechanism of semantic processing using knowledge graphs, graph theory
algorithms, semantic similarities and disambiguation. Our implementation can
be a relevant tool for studying the impact of events and users in the society.
Moreover the sentiment analysis in social networks can contribute in the public
health and design of custom services.
5 Acknowledgments
This work was supported by CONACYT and JSPS KAKENHI Grant Number
JP17H01789.
References
1. Princeton university "about wordnet." wordnet. princeton university (2010),
2. Ahuja, R.K., Mehlhorn, K., Orlin, J., Tarjan, R.E.: Faster algorithms for the shortest
path problem. Journal of the ACM (JACM) 37(2), 213223 (1990)
3. Baccianella, S., Esuli, A., Sebastiani, F.: Sentiwordnet 3.0: an enhanced lexical re-
source for sentiment analysis and opinion mining. In: LREC. vol. 10, pp. 22002204
(2010)
4. Bond, F., Baldwin, T., Fothergill, R., Uchimoto, K.: Japanese semcor: A sense-
tagged corpus of japanese. In: Proceedings of the 6th Global WordNet Conference
(GWC 2012). pp. 5663 (2012)
5. Bond, F., Foster, R.: Linking and extending an open multilingual wordnet. In: Pro-
ceedings of the 51st Annual Meeting of the Association for Computational Linguis-
tics (Volume 1: Long Papers). vol. 1, pp. 13521362 (2013)
6. Hwang, F., Richards, D.S.: Steiner tree problems. Networks 22(1), 5589 (1992)
7. Mohammad, S.M., Turney, P.D.: Crowdsourcing a wordemotion association lexi-
con. Computational Intelligence 29(3), 436465 (2013)