=Paper=
{{Paper
|id=Vol-2616/paper16
|storemode=property
|title=Model of Assessment of Information-Psychological Influence in Social Networking Services Based on Information Insurance
|pdfUrl=https://ceur-ws.org/Vol-2616/paper16.pdf
|volume=Vol-2616
|authors= Kateryna Molodetska, Yuriy Brodskiy, Solomiia Fedushko
|dblpUrl=https://dblp.org/rec/conf/coapsn/MolodetskaBF20
}}
==Model of Assessment of Information-Psychological Influence in Social Networking Services Based on Information Insurance==
https://ceur-ws.org/Vol-2616/paper16.pdf
Model of Assessment of Information-Psychological
Influence in Social Networking Services Based on
Information Insurance
Kateryna Molodetska1[0000-0001-9864-2463], and Yuriy Brodskiy2[0000-0002-6843-0192],
Solomia Fedushko3*[0000-0001-7548-5856]
1
Educational and scientific center of IT, Polissya National University,
Zhytomyr, Ukraine
2
Department of Computer Technologies and Systems Modeling, Polissya National University,
Zhytomyr, Ukraine
3
Lviv Polytechnic National University, 79013, Lviv, Ukraine
1
kmolodetska@gmail.com , yubrodskiy@gmail.com2
3
solomiia.s.fedushko@lpnu.ua
Abstract. In the current context, social internet services play a leading role in
various types of mass media. However, in the case of spreading destructive
content with the aim of informational-psychological influence on the actors,
social internet services can have a negative influence on the social and political
processes in the country. The article summarizes and systematizes the effects
created in virtual communities as a result of the implementation of threats to the
information security of the state. We have originated the model for evaluating
the informational-psychological impact on actors in the textual content of social
networking services based on conditional entropy. The model takes into account
not only the destructive informational-psychological impact from the content
source in social internet services but also the increase of the destructive
influence as a result of processing and further distribution of the content by
other actors of virtual communities. This approach allows increasing the
efficiency of detection of threats to the information security of the state in the
information space of social networking services. The developed model can be
used for the implementation of individual modules of the state information
security system in order to automate the detection of threats in the information
space and increase its efficiency.
Keywords: social networking service, virtual community, actor, information
security, informational-psychological influence, conditional entropy, text
content.
Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons
License Attribution 4.0 International (CC BY 4.0). COAPSN-2020: International Workshop on
Control, Optimisation and Analytical Processing of Social Networks
1 Introduction
The global development of information technology and the Internet have become
systemic factors in the development of the information society. As a consequence, it
has provoked changes in the system of strategic communication, where social
networking services (SNS) have become the most popular means of communication
[1-3]. Due to their high popularity, SNS can be used by the criminals to reach their
own goals in the information space of services. Manipulation of public opinion,
influence on the actors’ freedom of choice, their emotional and mental state,
discrediting the existing system of government in the state, etc. can be the result of
information and psychological influence on SNS actors [4-9]. Such phenomena are
widely spread not only in the national information space of Ukrainian SNS under the
conditions of information war with the Russian Federation but all over the world. For
instance, in January 2019, Russian media spread the information about Canada's
imposition of sanctions on Russia as a result of the seizure of power in Canada by
Ukrainian immigrants [10]. This is one of the numerous examples of a lengthy
information campaign held by Russian state media to misrepresent Ukraine and its
allies. Taking into account constant increase in the number of threats to the
information security of the state in SNS, related to the dissemination of manipulative
content, the problem of actualization of informational-psychological influence on
citizens in the information space of services seems especially important. The
development of technologies of latent information influence on the actors of virtual
communities, the lack of universal and effective methods of detecting destructive
influence determines the actuality of this work.
Analysis of recent research and publications [4, 11, 12–16] has shown that the
methods of manipulating of public consciousness of SNS actors are constantly
changing and improving. The existence of such a phenomenon is connected with the
continuous development of information-psychological methods of influence on the
actors in order to counteract its detection. In particular, the study [16] has found that
the following are currently the most popular means of manipulating SNS:
─ the use of labels in the posts of actors from the list of friends with whom the user is
not personally familiar. This allows the content to be distributed among the actor's
friends on his behalf;
─ the purchase of virtual communities of actors that are popular in the SNS
information space, ensuring that audience to be engaged in the spread of
propaganda and campaign materials;
─ blocking the profiles of actors, who are thought leaders, due to a large number of
bot complaints. Thus, the functioning of the actor’s page in the SNS is limited for
up to 30 days;
─ adding spam comments below the posts. That increases the size of the discussion
and leaves only a few comments that are often unrelated;
─ usage in the SNS of bots and trolls that have proven themselves to be the most
effective tool for conducting a hybrid war. They are also used for information
fighting at the level of state and local political groups;
─ generation of virtual communities with limited access that unite like-minded
people at the beginning of the creation and contribute to the effect of emotional
resonance between actors. Being created, such communities disseminate
destructive content that is fully perceived by its members which leads to
manipulation of public opinion.
─ usage of advertisement in SNS to spread unconcealed propaganda. Paid
publications do not pass the required level of verification to detect fakes. As a
result, it makes actors trust, capture, and distribute such content in the SNS.
Researchers [17] have also proved that the effects of provoking conflicts between
actors in SNS are monotony, ambivalence and desensitization. Monotony is
characterized by the usage of repeated calls, which makes actors stop responding to
them and lose interest in current affairs in the society and the state. This effect is
called “social fatigue”.
Ambivalence is an effect associated with a disruption of sensible news perception,
resulting in impulsive actions that are justified by encroachment on steady moral or
cultural values. Decreasing the level of sensitivity and empathy as a result of
ridiculing the manifestations of tolerance for violence or certain groups of the
population has been called the desensitization effect.
On the other hand, despite the active development of information technology and
psychological impact on actors in SNS, the development and improvement of
methods for detecting such influence are significantly limited. This is connected, in
particular, with the complexity of formalizing of the destructive content models and
lack of consistent features of the mentioned content. We have conducted the critical
analysis of publications devoted to the detection of manipulations in the information
space which has shown that articles [18–22] propose formal models of virtual
communities in SNS by describing them as an environment of information
confrontation with characteristics of the audience, public importance, content,
communication, state security allowing you to build a model of the management
system for the protection of virtual information space. However, the use of expert
polls to construct the models increases the level of the subjectivity of the results
obtained. The proposed approach does not take into account the change in the
characteristics of virtual communities with time. The article [23] proposes the method
of detecting public opinion manipulations based on the intellectual analysis and
consideration of information uncertainty in SNS text content. In this case,
unconditional entropy is an integral indicator of the threat to the information security
of the state in the SNS information space. However, increasing the efficiency of this
task can be achieved by using conditional entropy “IF condition THEN – an event”,
which is a promising area for further research.
The article aims to increase the effectiveness of detecting informational-
psychological impact on SNS actors based on conditional entropy, which will allow
taking into account not only the destructive informational impact from the source of
content but also the increase of such an impact due to its processing and further
distribution by other actors of virtual communities.
2 Models and methods
2.1 Content transformation with the informational-psychological impact as
means for manipulating public opinion
Content distributed in SNS is created directly by actors and characterized by a
diversity of forms and content. It reflects the subjective view of the author on events
in the country, society and the world. Let us investigate the peculiarities of identifying
signs of manipulation of public opinion in the textual content.
As researched in the publication [23], the connection between partial signs of
information and psychological influence in the SNS is represented in the form of a
hierarchy in a generalized form (Fig. [1]): F1 – presence of a reference to a subjective
point of view; F2 – lack of argumentation; F3 – share of interrogative sentences; F4
– doubtful statements; F5 – indicator of presence of exclamation sentences; F6 –
exclamations; F7 – adverbs; F8 – relative number of adverbs; F9 – usage of
emotional vocabulary; F10 – relative indicator of usage of the words to increase
attention; F11 – indicator of the usage of the words to emphasize the responsiveness
of events in content.
Integral feature
Partial features of
the first level
Partial features of
the second level
Fig. 1. Decision tree
According to the developed method [23], the signs F1 F5 are used to establish the
ambiguity of the facts Q1 stated in the text content of the SNS F6 F9 – its emotional
colouring Q2 , F10 F11 – manifestations of sensationalism Q4 . The presence of
features F1 F11 in the content is determined on the basis of calculating the frequency
of use of the given words-indicators. The hidden subject matter Q3 and tone is
connected with the usage of the methods of intellectual text analysis and machine
learning. This introduces an indicator of the informational entropy of SNS text
content, which determines the degree of uncertainty about the presence in the content
of hidden information influence on the actor. Information entropy is calculated on the
basis of the number of detected words-indicators. It decreases with increasing
frequency of their appearance and, accordingly, increases with low frequencies of
manifestation of the given words [24–26].
Thus, textual content created by the participants of informational operations
containing information-psychological influence on actors of virtual communities is
characterized by some amount of incoming entropy H ( x) or entropy of the content
source. After being distributed in the SNS information space, the content of the virtual
community actors is introduced. Then comes further distortion of the content due to
the peculiarities of its perception by the consumers and its subsequent publication
with the addition of subjective valuation judgments and conclusions.
Consequently, such content increases the level of information and psychological
impact on SNS actors H ( y / x) , which will be characterized by conditional entropy.
Then the loss of the meaning of the content after double distortion – first intentional
which involves the participants of informational operation, and then conducted by the
actors as the reaction to such content of directed meaning, is determined by the
amount of entropy H ( x / y) . The process is presented in Fig. [2] in the form of a
structural diagram.
content entropy
due to distortion
Undistorted content in the
information space
entropy of the SNS entropy of
content source end-user
loss of meaning of content
content
Fig. 2. The scheme of information processes in SNS in the conditions of manipulation of
public opinion
Undistorted content in the SNS I ( x, y) information space, which does not contain
information and psychological influence on the actors, can be formalized in the form
of the difference of the entropy of the content of the end consumer H ( y) and the
conditional entropy H ( y / x) , which characterizes the result of distorting the content
by the actors, or through the entropy of the content source H ( x) and conditional
entropy H ( x / y) which describes the loss of meaning of content as a result of its
transformation by actors
I ( x, y) H ( y) H ( y / x) H ( x) H ( x / y) . (1)
Therefore, the degree of uncertainty of the information-psychological impact on
actors in the SNS text content is determined by the unreliability of the process of
dissemination in the information space of virtual communities H ( x / y) or by the level
of additional distortion of the original content by the actors H ( y / x) .
2.2 Model of assessment of information-psychological influence
In general case, the scheme of interaction of actors, which is the source of content
X k , k 1, n in the SNS information space, and its consumers Yl , l 1, m is
presented in Fig. [3].
… …
Fig. 3. Generalized scheme of content exchange between SNS actors
For convenience, let us limit ourselves to the assumption that each content source in
the SNS and each consumer is connected by a network scheme, as shown in Fig. [4].
… …
Fig. 4. A partial case of connection between the k -source and the l -content consumer
From the publication [23] it is known that while processing a large number of
information messages in the SNS, the frequency of occurrence of the corresponding
words-indicators approaches the probability of occurrence of signs of information and
psychological impact. The model of such an influence should be described as a matrix
of transient probabilities. The matrix will be constituted from n 2 elements, where n -
the total number of signs of information and psychological influence on the actors.
Each element of the matrix, depending on the content of the estimated information
process in the SNS, represents a conditional probability p ( y j / xi ) or p ( xi / y j ) . The
matrix under consideration describes the transformation of content into an SNS that
spreads from the source X k to the consumer Yl and can take the following form
p ( y1 / x1 ) p ( y2 / x1 ) ... p ( yn / x1 )
p( y / x ) p ( y2 / x2 ) ... p ( yn / x2 )
p( y / x) 1 2
. (2)
... ... ... ...
p ( y1 / xn ) p ( y2 / xn ) ... p ( yn / xn )
Note that the sum of the elements of each row is equal to one
n n n
j 1
p ( y j / x1 )
j 1
p ( y j / x2 ) p( y / x ) 1 .
j 1
j n (3)
If the matrix of transient probabilities is known, then the frequencies of occurrence
of signs f jy of informational-psychological influence on the actors in the SNS at the
output are explicitly determined by the frequencies of signs f i x in the content spread
from the source, taking into account the transition probability p ( y j / xi )
n
f jy f p( y / x ) ,
i 1
i
x
j i (4)
or expanded
f1 y f1x p ( y1 / x1 ) f 2x p ( y1 / x2 ) ... f nx p ( y1 / xn ),
f 2y f1x p ( y2 / x1 ) f 2x p ( y2 / x2 ) ... f nx p ( y2 / xn ),
(5)
...
f ny f1x p ( yn / x1 ) f 2x p ( yn / x2 ) ... f nx p ( yn / xn ).
Then the entropy of the content consumed by SNS actors can be written in a form
of Shannon formula
n
H ( y) f log f .
j 1
j
y
2 j
y
(6)
Substituting equation (4) into expression (6) we obtain
n n n
H ( y) f p( y / x ) log f p( y / x ) .
j 1 i 1
i
x
j i 2
i 1
i
x
j i (7)
The component of SNS content that determines the informational-psychological
impact on the actors of virtual communities is represented as an expression of
conditional entropy
n n
H ( y / x) p ( x) p( y / x ) log p( y / x ) .
i 1 j 1
i j i 2 j i (8)
After substituting expressions (7) and (8) in equation (1) we obtain
n n n
I ( x, y ) H ( y ) H ( y / x ) f p( y / x ) log f p( y / x )
i 1 j 1
i
x
j i 2
i 1
i
x
j i
(9)
n n
p ( x) p( y / x ) log p( y / x ).
i 1 j 1
i j i 2 j i
Assuming that the likelihood of content containing destructive information from
the original source in the SNS approaches the frequency of occurrence of relevant
signs of threat pi ( x) fi x , then expression (9) is simplified to the form
n n
n
I ( x, y )
i 1 j 1
f i x p ( y j / xi ) log 2 p ( y j / xi ) log 2
i 1
f i x p ( y j / xi ) .
(10)
To interpret the numerical values obtained by expression (10), we use normalized
values I norm for information about the presence or absence of informational-
psychological influence on actors in the content of the SNS
I ( x, y )
I norm ( x, y) . (11)
I max ( x, y)
where I max ( x, y) is the maximum value of information.
Thus, the identification of informational-psychological influence on actors in the
SNS can be determined by the proportion of undistorted content I ( x, y) in the
information space of virtual communities, which is compared to the permissible limit.
The qualitative scale of threat information assessment of the psychological impact on
actors in SNS was formed as a result of the computational experiment and
generalization and adaptation of approaches to threat assessment in the field of
informational security (Table [1]) [23, 28].
Table 1. Adapted interval scale
Threat class Interval values of undistorted content I norm ( x, y )
very high 0,00–0,20
high 0,21–0,49
considerable 0,50–0,74
low 0,75–0,90
very low 0,91–1,00
3 Experiments
The textual content of the social network Facebook, as well as methods of interaction
with the API FB service and the integrated development environment of MS Visual
Studio have been used to evaluate the informational-psychological impact on the
actors.
Determining the tonality of textual content is realized on the basis of the Bayesian
multinomial naive method. Accordingly, the detection of hidden topics is realized
through the usage of probabilistic latent-semantic indexing. As a result of calculating
the entropy of partial signs of informational-psychological influence on public
opinion, the SNS obtained the following numerical values given in Table [2].
Table 2. The calculated values of entropy
HQ1 HQ2 HQ3 HQ4 HQ5
Value 0,30 0,40 0,34 0,52 0,50
A matrix P Y / X describing the transformation of the studied content, which is
spread from source X k to the consumer Yl in the SNS, where n 5 , takes the
following form
0,930 0, 010 0, 020 0, 030 0, 010
0, 010 0,944 0, 002 0, 014 0, 030
P Y / X 0, 040 0, 050 0,840 0, 050 0, 020 ,
0, 010 0, 020 0, 030 0,930 0, 010
0, 010 0, 030 0, 040 0, 020 0,900
and the frequency of appearance of signs f jy , j 5 informational-psychological
influence on the actors after its distribution in the information space and distortion are
equal to
f jy 0,3071 0, 4243 0,3299 0,5263 0, 4771 .
According to the expression (9), information concerning the presence or absence of
informational-psychological influence is equal to
I ( x, y) 2,5724 1, 2165 1,3559 , bit/feature.
The maximum value of information I max x, y in the absence of distortion of
content H y / x 0 acquires the value of 2.5734 bits/feature. Then the normalized
value of the information is I norm x, y 0,5271 . Thus, the level of threat to the
informational security of the state in the textual content under study is characterized
by a level of “significant”, which requires measures to counteract such destructive
information influence on the actors in the SNS [4, 12].
4 Conclusions
We have developed the model for evaluating the informational-psychological impact
on SNS actors in text content. The proposed approach uses the signs of manipulation
of actors in the information space of virtual communities. The mentioned signs are
summarized in the following such groups: doubtfulness of the presented facts;
emotional colouring; the presence of hidden topics; sensationality and tone.
The suggested estimation model differs from the known by the emergence of
information uncertainty, which is formalized as conditional entropy. This approach
takes into account the informational-psychological impact incorporated in the text by
the content creator, and the additional destructive impact of distorting that content
when processing and disseminating it to other actors in virtual communities. The
proposed model of assessment of informational-psychological impact provides the
increasing the efficiency of monitoring the information space of SNS and identifying
threats to the information security of the state. Thus, a general increase of the
efficiency and speed of the system providing information security of the state in the
SNS is achieved, which is an extremely urgent task for the modern society today.
References
1. Oxley, A.: Security risks in social media technologies: Safe practices in public service
applications. Elsevier (2013).
2. Li, Na, Sajal K. Das.: Efficiently discovering users connectivity with local information in
online social networks. Online Social Networks and Media 16, p. 100062 (2020).
doi:10.1016/j.osnem.2020.100062.
3. Haddow, G. D., Haddow, K. S.: Disaster communications in a changing media world.
Butterworth-Heinemann (2013).
4. Hryshchuk, R. Molodetska-Hrynhchuk, K.: Methodological foundation of state’s
information security in social networking services in conditions of hybrid war.
Information & Security: An International Journal 41, pp. 55–73 (2018).
5. Trach, O., Peleshchyshyn, A.: Functional-network model of tasks performance of the
virtual communities life cycle directions. In: Proceedings of the 11th International
Scientific and Technical Conference, CSIT 2016, pp. 108–110 (2016) doi:10.1109/STC-
CSIT.2016.7589881
6. Peleshchyshyn, A., Bandrovskyi, H.: Informational influence in social networks:
dynamics modeling based on the system of linear equations. In: Proceedings of 2019
IEEE 14th International Conference on Computer Sciences and Information
Technologies, CSIT Vol. 1, pp. 165–168 (2019, September) doi: 10.1109/STC-
CSIT.2019.8929819.
7. Korzh R., Peleshchyshyn A., Fedushko S., Syerov Y.: Protection of University
Information Image from Focused Aggressive Actions. Advances in Intelligent Systems
and Computing: Recent Advances in Systems, Control and Information Technology,
Proceedings of the International Conference SCIT 2016, vol. 543, pp. 104-110. Warsaw,
Poland. Springer (2017).https://doi.org/10.1007/978-3-319-48923-0_14
8. Korobiichuk, I., Snitsarenko, P., Katsalap, V., Hryshchuk, R.: Determination and
Evaluation of Negative Informational and Psychological Influence on the Military
Personnel Based on the Quantitative Measure. In: Proceedings of the 1st International
Workshop on Control, Optimisation and Analytical Processing of Social Networks
(COAPSN-2019) 2392 pp. 66–78 (2019).
9. Peleschyshyn, A., Holub, Z., Holub, I.: Methods of real-time detecting manipulation in
online communities. In: Computer Sciences and Information Technologies - Proceedings
of the 11th International Scientific and Technical Conference, CSIT 2016, pp. 15–17
(2016). doi:10.1109/STC-CSIT.2016.7589857
10. MediaSapiens: https://ms.detector.media/ethics/manipulation
11. Hryshchuk, R., Molodetska, K., Syerov, Y.: Method of improving the information
security of virtual communities in social networking services. In: CEUR Workshop
Proceedings 2392 pp. 23–41 (2019)
12. Hryshchuk, R., Molodetska, K.: Synergetic control of social networking services actors’
interactions. In: Recent Advances in Systems, Control and Information Technology 543
pp. 34–42 (2017) doi:10.1007/978-3-319-48923-0_5
13. Cheng, Chun, Yun Luo, and Changbin Yu: Dynamic mechanism of social bots interfering
with public opinion in network. Physica A: Statistical Mechanics and its Applications
p. 124163 (2020) doi:10.1016/j.physa.2020.124163.
14. Etesami, S., Rasoul, et al.: Evolution of Public Opinion under Conformist and
Manipulative Behaviors. IFAC-PapersOnLine 50.1 pp. 14344–14349 (2017) doi:
10.1016/j.ifacol.2017.08.2012.
15. Luwei Rose L., Schmierbach M., Yu-Leung Ng.: Willingness to follow opinion leaders:
A case study of Chinese Weibo. Computers in Human Behavior 101 pp. 42–50 (2019)
doi: 10.1016/j.chb.2019.07.005.
16. Independent publishing TEXTY.org.ua (in Ukrainian)
http://texty.org.ua/pg/article/editorial/read/85655/Tehnika_manipulaciji_Visim_brudnyh_
pryjomiv_jaki_vykorystovujut.
17. Virtual warfare: how social networks are used to stir up conflict:
https://konkurent.in.ua/publication/40122/virtualna-viyna-yak-vikoristovuut-sotsmerezhi-
dlya-rozpaluvannya-konfliktiv/.
18. Kereselidze, N.: Combined continuous nonlinear mathematical and computer models of
the information warfare. International Journal of Circuits, Systems and Signal Processing
12, pp. 220–228 (2018)
19. Molodetska K., Tymonin Yu., Melnychuk I.: The conceptual model of information
confrontation of virtual communities in social networking services. International Journal
of Electrical and Computer Engineering 10(1) pp. 1043–1052 (2020).
20. Hryshchuk R., Molodetska K., Tymonin Yu.: Modelling of conflict interaction of virtual
communities in social networking services on an example of anti-vaccination movement.
In: Proc. of the International Workshop on Conflict Management in Global Information
Networks CMiGIN 2019 2588 pp. 250–264 (2019).
21. Zakharchenko A., Maksimtsova Y., Iurchenko V., Shevchenko V., Fedushko S.: Under
the conditions of non-agenda ownership: Social media users in the 2019 Ukrainian
presidential elections campaign. CEUR Workshop Proceedings, vol. 2392, pp. 199-219.
Lviv, Ukraine (2019).
22. Syerov Yu., Fedushko S., Loboda Z.: Determination of Development Scenarios of the
Educational Web Forum. Proceedings of the XIth International Scientific and Technical
Conference (CSIT 2016), pp. 73-76. Lviv (2016). https://doi.org/10.1109/STC-
CSIT.2016.7589872
23. Molodetska-Hrynchuk, K.: Methodology of detection of the public opinion manipulations
in the social networking services (in Ukrainian). Information security 4(24) 80–92
(2016).
24. Thuy, N. N., Wongthanavasu, S.: On reduction of attributes in inconsistent decision
tables based on information entropies and stripped quotient sets. Expert Systems with
Applications 137 pp. 308–323 (2019) doi: 10.1016/j.eswa.2019.06.071.
25. Wang, Y., Yongming, L.: Bayesian entropy network for fusion of different types of
information. Reliability Engineering & System Safety 195 106747 (2020) doi:
10.1016/j.ress.2019.106747.
26. Li, DongDong, et al.: Information entropy based sample reduction for support vector data
description. Applied Soft Computing 71 pp. 1153–1160 (2018) doi:
10.1016/j.asoc.2018.02.053.
27. Dai, Jianhua, et al.: Uncertainty measurement for interval-valued decision systems based
on extended conditional entropy. Knowledge-Based Systems 27 pp. 443–450 (2012) doi:
10.1016/j.knosys.2011.10.013.
28. Hryschuk, R., Mamarev, V.: Optimization method of the input data flow dimensionality
for the information protection systems (in Ukrainian). Information security 2(8) pp. 27–
34 (2012).