Social media influence analysis Techniques Systematic Literature Review

Abstract

Nowadays, the use of Social Media networks is growing endlessly and rapidly, those networks have become a substantial pool for unstructured data. Social media influence (SMI) describes the social media influencers (SMIs) capacity to influence other people's thinking, feelings and characteristics in online and outline communities. The analysis of the influence activity afects all diferent kinds of ifelds from multiple perspectives such as strategic planning and decision making until product creation and distribution. The main contribution of this paper is to present results of a Systematic Literature Review (SLR) that highlights the diferent Techniques used in the analysis of social media influence, when addressing influencers-followers interactions. After a careful review of the 55 extracted articles, we found that 4 data representation models have been used with social media related data analysis and 10 data analysis techniques to address 6 diferent research objectives in more than 20 diferent fields. In Interactions and users relationships purpose, Graph was the most used data representation model and data analysis techniques.

Keywords

• eol>Influence
• Systematic literature review (SLR)
• Social media influence (SMI)
• Data analysis techniques

1. Introduction

According to the Statista report [ 1 ], over 3.6 billion people were using social media worldwide, a number projected to increase to almost 4.41 billion in 2025. The report shows that 4.57 billion people around the world use the internet, of those users, 346 million new users have come online within the last 12 months. Internet users spend an average of 144 minutes on social media per day. The process of analyzing or mining social networks helps in gathering information that optimize influence maximization. People use Social Media Platforms to connect with their friends and family members, to introduce themselves to others by sharing their daily live news and follow channels or pages. This spontaneous behavior created what we now call influencers and followers. Which change the way that organisations connect with their clients. Our Study helps in choosing the appropriate representation model and analysis technique that matches the analysis purpose and goes with certain platforms. The better way to answer a questions of efectiveness comparing more than one diferent bath is Systematic Literature Reviews. This paper is organized as follows. In Section 2, we represent the Systematic Literature Review planning. Results and Discussion are presented in Section 3. We conclude with general highlights and perspectives for further work in Section 4.

2. Systematic Literature Review Planning

2.1. Research Questions

Aiming to find all relevant primary studies related to the diferent types of models used to describe the phenomenon of influence in social media networks, the following research questions (RQ) were established: • RQ1: Which techniques have been used to analyse users related data in social media? • RQ2: What is the purpose of analyzing users related data in social media? • RQ3: How is it possible to use analyzing techniques when addressing influencers/followers interactions? Subsequently, we determined the initial research in the databases. In relation to the keywords, three groups were formed: • Groupe1: (“social media",“social networks") • Groupe2: (“analysis", “analytics", “analyzing", “analyze", “content analysis") • Groupe3: (“influencers",“influencer", “followers" ,“follower")

2.2. Search Strategy

The search strategy combines the key concepts of our search question in order to retrieve accurate results. It is an organized structure of key words, which are “social media”, “analysis”, “influencer” and “followers”, used to search a database. Then, we added synonyms, variations, and related terms for each keyword. A Boolean operator (AND and OR) allow us to try diferent combinations of search terms.The final search string is (“social media” OR “social networks”) AND (“analysis” OR “analytics” OR “analyzing” OR “analyze” OR “content analysis”) AND (“influencers” OR “influencer” OR “followers” OR “follower”).

2.3. Selection Criteria

After obtaining the search results from the diferent sources, a set of exclusion/ inclusion criteria was applied to help in the identification of relevant primary studies. Therefore, Inclusion Criteria (IC) are used to select primary studies which indicate Related data analysis techniques, purpose, or influencers/followers interactions for Social media networks. For the Exclusion Criteria (EC) they are used to remove those primary studies that do not address the main topics searched in this SLR, are not available, or are directly related to an included primary study of the same author. • Inclusion Criteria (IC): – Publications that match one of the search items – Publications that have best practices version – Publications that are related to social media networks related data analysis – Publications that are related to the phenomenon of influence in social media networks – Publications that are relate to the research questions • Exclusion Criteria (EC): – Publications that not match one of the search items – Publications that do not have best practices version – Publications that are published before or on the 31.12.1999 – Publications that are not related to the phenomenon of influence in social media networks – Publications that are not relate to the research questions

2.4. Data collection

The number of papers resulting in the search is summarized in TABLE 1. After filtering irrelevant, duplicate and incomplete papers, a total of 55 papers in TABLE 3 were selected for the reviewing process. TABLE 2 presents the filtering process. The state of the art is presented as follows according to the diferent cases. The selected papers per resources are distributed as shown in TABLE 3.

3. Results and Discussion

3.1. Social Media Users Related data analysis techniques

3.1.1. Data model: When we start defining the Social Media Users Related data analysis techniques we found that the first thing we need to see is the real-state of the abstract data or the data model that defines the initial information. There are diferent models to represent it, we chose to classify them into 4 main categories summarized in TABLE 4.

Data model

Graphs Dataset

Log Diagrams

Numbers of Papers 33 9 7 6

Percentage 73.33%

20% 15.45 % 13.33%

Graphs: are the most common used data model. Each Graph has his own parameters related to the topic and the field of search. The social networks is the top sous-category with more than 15 articles [ 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18 ], it defines a social structure made up from a set of diferent social actors. Some authors create a specific model such as Small-world network multiple influence model (SWMI model) [ 19], Community-Author-Recipient-Topic model (CART) [20], RT and MT model [ 4 ]. There are some types of Networks related only to influence analysis like User-Follower Network [ 21], influencer Graph [ 22]. The general ones are Graph Neural Networks (KOLs), Knowledge Graph(KG) [23], Random geometric graph, random networks [24] and Network diagram [25]. Peng, Sara, Taeho, Hui, Krishna, Xiang-Yang, Kevin and al. [ 26, 9 ] used Social Graph that represents social relations between entities. Some authors mix between more then one representation. Marco and Mattia [27] used Graph Representation Learning. Lauren, Robert, Augustin and Eugene [ 3 ] used network diagram of operation model and Graph-based representation for trust/reputation systems. Fan and Cassandra [28] used Onemode and two-mode network, interaction networks and class-level and group-level discussion networks. Mozhgan and Kevin C. [29] used Graph-based data representation and Hypergraph data representation.

Datasets: are the second category, each platform has a specific type of data. Twitter dataset [30] is compiled from various tweets which centered on topics, hashtags, and objects. Epinions dataset [31, 32] is the organization of data into incremental snapshots. Facebook dataset [33] is more complicated, representing the number of acting users, number of users that reacted, number of posts, number of comments and time span of data. Some datasets are related to the real-world for example to study the Location-based social networks (LBSN) [34].

Logs: category has 3 sub-category. Venkata, Weizhong and Xiaowei [21] used logarithms in comparing the results founded in the Twitter User-Follower network follows power-law degree distribution. Also, Simone, Diego, Giuseppe and Maurizio [35] used logarithms in his study. The second sub-category is Big data which refers to the large, fast or complex type of data that it’s dificult or impossible to process using the traditional methods. There are four papers mention this type [12, 36, 37, 38]. The last sub-category is Clustering [13], it is used to extract the Trusted and Non-Trusted nodes.

Diagrams: category results to a schematic representation. Mozhgan and Kevin C. [29] used Fuzzy models and diagrams that look like Graph. Sunagul [ 5 ] used Sociogram, it is a graphic representation of social links that a person has. Also Lars and Francis applied Sociogram that represent participant’s friendship network. Marco and Mattia [39] used schematic diagram using social relation. 3.1.2. Data Analysis model In this part we will focus on the diferent data analysis model used to analyse Social Media Users Related data. We propose 10 main categories represented in TABLE 5. Probabilistic model and Graphs are the earliest used analysis techniques. Every work ad at least one of the three statistical categories: tables, curves and histograms. The most used category is Analysis model. The latest work used mining techniques like Artificial intelligence (AI) and Prediction.

Data Analysis model

Analysis model

Tables

Curves

Ranking metrics Artificial intelligence(AI)

Graphs Histograms

Prediction Probabilistic model numeric coeficient

Analysis model: is a technical representation that results from designing a model that analyse diferent information, behavior, function, interaction or relations. The ones related to influence are: Influence model [ 40, 37, 11 ] Social media Influencer Model[ 25], Action-Reaction Influence Model (ARIM) [ 33, 8 ], Influence evaluation model [ 2 ] and Influence model for sentiment inference [ 10 ]. Others are related to topics that helps in decisions making such as recommendation model [36], Latent Dirichlet Allocation topic model [41], Graphical representation of Latent Dirichlet Allocation [26], Multi-Criteria Decision Making model (MCDM) [37] and numerous theoretical models [ 30, 11, 38 ]. Other types are Difusion models like Flow diagrams and Block diagrams [29, 26, 22], Tow-step-flow model, multi-step-flow model [ 4 ], Filtering model [30, 38] and Emerging Model [42]. Also there is models related to social or marketing analysis: The inbound and outbound models [43], Alternative Communications Model theory [44], Covariance-based structural equation modeling (CB-SEM) [42] hypothesized mediation model [17] and Social Information Retrieval [45].

Tables: are used to represent comparisons e.g. Country-Level Micro-blog User Behaviour and Activity [39], Models on top-K Recommendation [23] Top 10 users diferent diagrams [ 37], Direct Trust in communities and network [13], classifiers [ 35] and probability distributions by groups [46]. Also, to describe number of users, location, visits and clustered locations[34]. To present results like Words frequency[38], Linear Regression Results [ 9 ] Kendall Rank Correlation, Hierarchical Multiple Regression [12]. Finally, to list results like randomly picked communities from the observed communities [21].

Curves: are used to represent correlations in most of them e.g. correlation strength between diferent countries [ 39], Pearson’s Correlation [47], descriptive statistics of key study variables [14], Descriptive correlations [17] and Correlation between user features [48]. There are analysis models used to extract those representation like: Principal component analysis [49], Difusion models [26], Bilinear state space model (BSSM) [ 11 ], Empirical model [50], Model R, model R-Squared, Adjusted R-Squared [15], Bass model [19] and Opinion dynamics models [18]. They are also used to represent evaluations like: Average week-on-week growth rates, Social media followers by week [50] and Evolution of retweeting network [48] or to represent Influence probability [34]. Sara, Taeho, Hui, Krishna, Xiang-Yang and Kevin [ 9 ] used Life Curve. Saike, Xiaolong, Danie, Kainan, Zhu and Chuan [ 11 ] used ROC curve.

Histograms: are used to represent quantity evolution throw time e.g. Histogram of the Influence Score [ 9 ], Absolute and Relative influence [ 34], account creation dates for Twitter followers of incumbent US senators campaigning in 2018 [51] and probability of influence relationship [ 10 ]. Also to show Participant centralities for diferent networks [28].

Ranking metrics: are an algorithms used to rank components of social media like tweets [39, 12, 46], posts, Hashtags, Trends [52], Page rank [ 8 ] or even influencers and followers ranking. They are often used to analyse users related data in Twitter. Also we can use them to rate parameters like frequency and percentage e.g. the frequencies for the component items of Twitter and YouTube use [15], The percentage of users following at least one of the top (key) opinion leaders [23], Percentage distribution of top influencers [ 41]. The most used techniques to characterize top users are ranking metrics like Swiss Market Index (SMI) [ 4 ] and NavigTweet [53]. This technique is also used to realize comparisons between the most powerful influencers according to betweenness centrality and Page Rank and worth mentioning the hashtags [ 5 ].

Artificial intelligence(AI): analysis techniques has started in recent years. There are 5 sup-categories: (1) Clustering e.g. K-means clustering method [27], Distribution of nearest neighbors of regional network nodes [39] and K-means clustering method [39]. (2) Machine learning techniques [30, 38]. (3) Heuristic model e.g. energy-propagation model [ 9 ], influence propagation model [ 2 ] and difusion model based on cascade model [ 48]. (4) Mining techniques like Mining Micro-Influencers [ 35]. (5) knowledge engineering e.g. Knowledge representation and reasoning [39].

Graph: are not only used for primary data extraction but also for deep learning methods to describe data by graphs designed. There are a multitude of model to design graph such as: Independent Cascade Model [ 44, 2 ], Linear Threshold Model [ 9 ], Heterogeneous Influence graph model [30] and Trust Model (SNTrust) [13]. Also there are a diferent types of networks such as: Bayesian network [29], Two-link network topology, Parallel link topology [24] and social network [44]. Those models also used to extract Social Network graph density reduction [20] and Instructor’s centralities [28].

Prediction: are models used to give a future vision or estimation. The most used prediction modelis the Standing Ovation model (SOM) [30, 38]. Daekook, Bomi, Byoungun, Youngjo and Yongtae [19] creates more then one estimation to compare between them. Paul, Liam and Jordi [50] used diference-in-diference models of social media followers to analyse the social media music fans followers future behavior.

Probabilistic model: in the field of social media analysis are used for diferent purposes: (1) mining structural influence to analyze relationships in social network [ 47], (2) identification of influencers in online social networks [ 20], (3) analyzing dynamics of information difusion [ 48], (4) evaluating Role of Conformity in Opinion Dynamics in Social Networks [18], (5) modeling Topic [44].

Numeric coeficient: are used for diferent reasons one of them is to represent the size of an individual’s social network and their ability to influence that network [ 40]. Also, to show how influence scores change [ 6 ]. Some of the authors use numeric coeficient to rate the most frequent words and User quality ratio vs. RT Quality ratio vs. Replay ratio for the Top users [30]. Pearson Coeficient [54] is one of the most used coeficients.

3.2. Social Media Users Related data analysis purpose

Reading all the 55 articles we found six main important reasons presented as follow in TABLE 6 behind analyzing users related data.

Analysis main purpose Interactions and users relationships

Influencers Behaviour

Influence Modeling Influence Evaluation

Mining Influence Influence Optimisation

Numbers of Papers 13 13 10 9 6 4

Percentage 23.64% 23.64% 18.18 % 16.36% 10.91% 7.27%

3.3. Social Media Users Related data analysis search Fields

After deduce the six main reasons, we found that each of them is related with a search field or maybe more than one. While reading other SLR related to the social media analysis they frequently mention that there is an extensive variety of fields benefit from social media related data analysis, but most of them chose to focus between one, two and three domains. Based on 3.0 Detailed (four digit) subject codes, we extract 23 fields deduct from 4 main fields: (1) Humanities and social science, (2) Natural sciences, (3) Formal sciences and (4) Professions and applied sciences, more than 60 sub-field and more than 100 sub-sub-field. We notice that it afects and optimises all diferent kinds main fields. The four top fields are Sociology with 41.81%, Business with 36.36%, Computer sciences 30.90% and Interdisciplinary studies with 27.26%.

3.4. Social Media Users interactions analysis techniques

After extracting the 6 main purposes of analysing social media users related data we found that the top of them is: Interactions and users relationships with a total of 13 articles (bibliographic portfolio). More than 61% of those papers used Graphs as a representation, 23% used Dataset, 15% used Logs and 8% used Diagrams. Coming to the analysis techniques most of the authors prefer to combine between more than one techniques. Nadia, Mourad, Lin, Ben, Yousra, Ahmad Kamran, Basit, Ahmad Raza, Fan and Cassandra used Graphs [ 10, 47, 13, 28 ]. Monika, Amel, Katarzyna, Alda, Nadia, Mourad, Lin and Ben [ 33, 10, 47 ] used analysis Model. Marco, Mattia, Lauren, Robert, Augustin and Eugene [ 27, 3 ] used Ranking Metrics. For the statistic techniques: Yousra, Ahmad Kamran, Basit, Ahmad Raza, Venkata, Swamy, Weizhong and Xiaowei [13, 21] used tables. Yousra, Ahmad Kamran, Basit, Ahmad Raza, Lin, Ben, Faisal M., Ramaravind and Joyojeet [54, 47, 13] used curves. Nadia, Mourad, Fan and Cassandra [ 10, 28 ] used Histograms. Marco and Mattia [27] added Artificial Intelligence, Lin and Ben [ 47] added Probabilistic Model and Prediction also Faisal M., Ramaravind and Joyojeet [54] added Numeric Coeficients. The most used platforms for interactions analysis are Twitter and Facebook, but in 2019 researchers are concentrated more on Instagram. For the fields there are a diversity in the chosen fields but the most figured ones were Sociology and Business.

4. Conclusion

The present Systematic Literature Review sought to contribute in the identification of users related data analysis techniques as well as to present the analysis main purpose and the fields that were most engaged regarding social media influence. This review consisted of literature published between 2000 and 2020. We did list 4 data representation models, 10 data analysis techniques, 6 main purposes behind social media related data analysis and more than 20 fields were extracted. The most used analysis techniques were Graphs as for the data representation model. The most-frequently appeared purpose was interactions and users relationships analysis. For the fields we found that Sociology, Business and Computer sciences are the top connected ifelds with social media industry. As a future work, we plan to develop a mining approach to extract the interaction model between influencers and followers, using graphs. Further more we will apply graph matching and transformation techniques on the interaction model, using GMTE, for analysis purposes.

Acknowledgments

This work was partially supported by the LABEX-TA project MeFoGL:"Méthodes Formelles pour le Génie Logiciel". social networks using transfer entropy, in: Pacific-Asia Workshop on Intelligence and Security Informatics, Springer, 2013, pp. 47–61. [12] X. Wang, Y. Yu, L. Lin, Tweeting the united nations climate change conference in paris (cop21): An analysis of a social network and factors determining the network influence, Online Social Networks and Media 15 (2020) 100059. [13] Y. Asim, A. K. Malik, B. Raza, A. R. Shahid, A trust model for analysis of trust, influence and their relationship in social network communities, Telematics and Informatics 36 (2019) 94–116. [14] K. J. Forney, T. Schwendler, R. M. Ward, Examining similarities in eating pathology, negative afect, and perfectionism among peers: A social network analysis, Appetite 137 (2019) 236–243. [15] A. Ioanid, C. Scarlat, Factors influencing social networks use for business: Twitter and youtube analysis, Procedia Engineering 181 (2017) 977–983. [16] O. E. Llantos, M. R. J. E. Estuar, Characterizing instructional leader interactions in a social learning management system using social network analysis, Procedia Computer Science 160 (2019) 149–156. [17] N. Kwok, S. Hanig, D. J. Brown, W. Shen, How leader role identity influences the process of leader emergence: A social network analysis, The Leadership Quarterly 29 (2018) 648–662. [18] A. Das, S. Gollapudi, A. Khan, R. Paes Leme, Role of conformity in opinion dynamics in social networks, in: Proceedings of the second ACM conference on Online social networks, 2014, pp. 25–36. [19] D. Kang, B. Song, B. Yoon, Y. Lee, Y. Park, Difusion pattern analysis for social networking sites using small-world network multiple influence model, Technological Forecasting and Social Change 95 (2015) 73–86. [20] S. A. Rios, F. Aguilera, J. D. Nuñez-Gonzalez, M. Graña, Semantically enhanced network analysis for influencer identification in online social networks, Neurocomputing 326 (2019) 71–81. [21] V. Martha, W. Zhao, X. Xu, A study on twitter user-follower network: A network based analysis, in: Proceedings of the 2013 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM ’13, Association for Computing Machinery, New York, NY, USA, 2013, p. 1405–1409. [22] E. D.-j. Kim, B. J.-l. Keng, K. Padmanabhan, Systems and methods for dynamically determining influencers in a social data network using weighted analysis, 2016. US Patent 9,262,537. [23] J. Wang, K. Ding, Z. Zhu, Y. Zhang, J. Caverlee, Key opinion leaders in recommendation systems: opinion elicitation and difusion, in: Proceedings of the 13th International Conference on Web Search and Data Mining, 2020, pp. 636–644. [24] J. Elias, F. Martignon, K. Avrachenkov, G. Neglia, A game theoretic analysis of network design with socially-aware users, Computer Networks 55 (2011) 106–118. [25] O. Okuah, B. M. Scholtz, B. Snow, A grounded theory analysis of the techniques used by social media influencers and their potential for influencing the public regarding environmental awareness, in: Proceedings of the South African Institute of Computer Scientists and Information Technologists 2019, 2019, pp. 1–10. [26] W. Peng, T. Sun, Method and system for identifying a key influencer in social media utilizing topic modeling and social difusion analysis, 2012. US Patent 8,312,056. [27] M. Brambilla, M. Gasparini, Brand community analysis on social networks using graph representation learning, in: Proceedings of the 34th ACM/SIGAPP Symposium on Applied Computing, 2019, pp. 2060–2069. [28] F. Ouyang, C. Scharber, The influences of an experienced instructor’s discussion design and facilitation on an online learning community development: A social network analysis study, The Internet and Higher Education 35 (2017) 34–47. [29] M. Tavakolifard, K. C. Almeroth, Social computing: an intersection of recommender systems, trust/reputation systems, and social networks, IEEE Network 26 (2012) 53–58. [30] M. Al-Qurishi, S. Alhuzami, M. AlRubaian, M. S. Hossain, A. Alamri, M. A. Rahman, User profiling for big social media data using standing ovation model, Multimedia Tools and Applications 77 (2018) 11179–11201. [31] M. Paquet-Clouston, O. Bilodeau, D. Décary-Hétu, Can we trust social media data? social network manipulation by an iot botnet, in: Proceedings of the 8th International Conference on Social Media & Society, Association for Computing Machinery, New York, NY, USA, 2017. [32] S. Agreste, P. De Meo, E. Ferrara, S. Piccolo, A. Provetti, Trust networks: Topology, dynamics, and measurements, IEEE Internet Computing 19 (2015) 26–35. [33] M. E. Rakoczy, A. Bouzeghoub, K. Wegrzyn-Wolska, A. L. Gancarski, Exploring interactions in social networks for influence discovery, in: International Conference on Business Information Systems, Springer, 2019, pp. 23–37. [34] M. A. Saleem, R. Kumar, T. Calders, T. B. Pedersen, Efective and eficient location influence mining in location-based social networks, Knowledge and Information Systems 61 (2019) 327–362. [35] S. Leonardi, D. Monti, G. Rizzo, M. Morisio, Mining micro-influencers from social media posts, in: Proceedings of the 35th Annual ACM Symposium on Applied Computing, SAC ’20, Association for Computing Machinery, New York, NY, USA, 2020, p. 867–874. [36] X. Li, X. Yan, Analyzing of personalized recommendation model of social network users based on big data, in: 2019 6th International Conference on Dependable Systems and Their Applications (DSA), IEEE, 2020, pp. 135–142. [37] A. Muruganantham, G. M. Gandhi, Framework for social media analytics based on multicriteria decision making (mcdm) model, Multimedia Tools and Applications 79 (2020) 3913–3927. [38] X. Yang, S. Kim, Y. Sun, How do influencers mention brands in social media? sponsorship prediction of instagram posts, in: Proceedings of the 2019 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM ’19, Association for Computing Machinery, New York, NY, USA, 2019, p. 101–104. [39] P. Yang, J. Liu, J. Qi, Y. Yang, X. Wang, Z. Lv, Comparison and modelling of country-level microblog user and activity in cyber-physical-social systems using weibo and twitter data, ACM Transactions on Intelligent Systems and Technology (TIST) 10 (2019) 1–24. [40] I. Anger, C. Kittl, Measuring influence on twitter, in: Proceedings of the 11th international conference on knowledge management and knowledge technologies, 2011, pp. 1–4. [41] N. Li, D. Gillet, Identifying influential scholars in academic social media platforms, in: Proceedings of the 2013 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, 2013, pp. 608–614. [42] M. C. O. Mariano, J. C. M. Maniego, H. L. M. D. Manila, R. C. C. Mapanoo, K. M. A. Maquiran, J. R. B. Macindo, L. M. S. Tejero, G. C. S. Torres, Social media use profile, social skills, and nurse-patient interaction among registered nurses in tertiary hospitals: a structural equation model analysis, International journal of nursing studies 80 (2018) 76–82. [43] D. Hatcher, G. S. Bawa, B. de Ville, How you can identify influencers in sas r social media analysis (and why it matters), in: SAS Global Forum, Citeseer, 2011, pp. 4–7. [44] M. van Staden, L. van Niekerk, Uncovering the value of influencer marketing through social network analysis and brand positioning insights, SAMRA,[Online] Available from: http://www. samra. co. za/wp-content/uploads/Van-Staden-Van-Niekerk_Uncovering-thevalue-of-influencer-marketing. pdf (2018). [45] P. S. Ludu, Inferring latent attributes of an indian twitter user using celebrities and class influencers, in: Proceedings of the 1st ACM Workshop on Social Media World Sensors, 2015, pp. 9–15. [46] S. Hong, D. Nadler, Social media and political voices of organized interest groups: A descriptive analysis, in: Proceedings of the 16th Annual International Conference on Digital Government Research, dg.o ’15, Association for Computing Machinery, New York, NY, USA, 2015, p. 210–216. [47] L. Guo, B. Zhang, Mining structural influence to analyze relationships in social network,

Physica A: Statistical Mechanics and Its Applications 523 (2019) 301–309. [48] Y. Zhou, B. Zhang, X. Sun, Q. Zheng, T. Liu, Analyzing and modeling dynamics of information difusion in microblogging social network, Journal of Network and Computer Applications 86 (2017) 92–102. [49] J.-V. Cossu, N. Dugué, V. Labatut, Detecting real-world influence through twitter, in: 2015

Second European Network Intelligence Conference, IEEE, 2015, pp. 83–90. [50] P. Crosby, L. J. Lenten, J. McKenzie, Social media followers as music fans: Analysis of a music poll event, Economics letters 168 (2018) 85–89. [51] R. Takacs, I. McCulloh, Dormant bots in social media: Twitter and the 2018 u.s. senate election, in: Proceedings of the 2019 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM ’19, Association for Computing Machinery, New York, NY, USA, 2019, p. 796–800. [52] C.-w. Shen, C.-J. Kuo, P. T. Minh Ly, Analysis of social media influencers and trends on online and mobile learning, International Review of Research in Open and Distributed Learning 18 (2017) 208–224. [53] C. Francalanci, A. Hussain, Navigtweet: A visual tool for influence-based twitter browsing, in: International Conference on Design Science Research in Information Systems, Springer, 2015, pp. 183–198. [54] F. M. Lalani, R. Kommiya Mothilal, J. Pal, The appeal of influencers to the social media outreach of indian politicians, in: Conference Companion Publication of the 2019 on Computer Supported Cooperative Work and Social Computing, 2019, pp. 267–271.

References

[1] H. Tankovska , Number of social network users worldwide from 2017 to 2025 (in billions ), 2017 . URL: https://www.statista.com/statistics/260811/ social-network -penetration-worldwide/.

[2] T. Leung , F.-l. Chung, Persuasion driven influence propagation in social networks , in: 2014 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM 2014 ), IEEE, 2014 , pp. 548 - 554 .

[3] L. Arnett , R. Netzorg , A. Chaintreau , E. Wu, Cross-platform interactions and popularity in the live-streaming community , in: Extended Abstracts of the 2019 CHI Conference on Human Factors in Computing Systems , 2019 , pp. 1 - 6 .

[4] M. del Fresno Garcia , A. J. Daly , S. Segado Sanchez-Cabezudo, Identifying the new influences in the internet era: Social media and social network analysis ., Revista Española de Investigaciones Sociológicas ( 2016 ).

[5] S. Sani-Bozkurt , Identifying network structure, influencers and social mood in digital spheres: A sentiment and content analysis of down syndrome awareness ., World Journal on Educational Technology: Current Issues 10 ( 2018 ) 10 - 19 .

[6] I. Garibay , A. V. Mantzaris , A. Rajabi , C. E. Taylor , Polarization in social media assists influencers to become more influential: analysis and two inoculation strategies , Scientific reports 9 ( 2019 ) 1 - 9 .

[7] L. Groeger , F. Buttle , Influencers and their circle of friends: enriching social network analysis with qualitative data , in: Australian and New Zealand Marketing Academy Conference ( 2012 ), ANZMAC2012 Conference, 2012 , pp. 1 - 7 .

[8] M. E. Rakoczy , A. Bouzeghoub , A. L. Gancarski , K. Wegrzyn-Wolska, In the search of quality influence on a small scale-micro-influencers discovery, in: OTM Confederated International Conferences" On the Move to Meaningful Internet Systems" , Springer, 2018 , pp. 138 - 153 .

[9] S. Motahari , T. Jung , H. Zang , K. Janakiraman , X.-Y. Li , K. S. Hoo , Predicting the influencers on wireless subscriber churn , in: 2014 IEEE Wireless Communications and Networking Conference (WCNC) , IEEE, 2014 , pp. 3402 - 3407 .

[10] N. Chouchani , M. Abed , Enhance sentiment analysis on social networks with social influence analytics , Journal of Ambient Intelligence and Humanized Computing 11 ( 2020 ) 139 - 149 .

[11] S. He , X. Zheng , D. Zeng , K. Cui , Z. Zhang , C. Luo, Identifying peer influence in online