6
Social Networks
For a decade, social networks have been unavoidable, whatever one’s opinion on the quality of the information they convey. Designed from the outset as a means of communication, of information sharing and a tool for creating communities who wish to converse and create, they have partly deviated from their original aim. In fact, they play a role in the ever-faster diffusion of information and because of this, they are capable of making millions of people aware of specific information in a very short space of time. We are therefore seeing what individuals, institutions, businesses and States with ill-intent can do with these “communication tools”. These networks are therefore playing a full role in influence operations, which is to say they aim – especially via “viral marketing” – to create the desired reaction in exposed individuals, without them being aware of the “manipulation” underway.
Moreover, more and more “socialbots” are able to penetrate social networks and steal the identities of some individuals and, because of this, to provide social networks with “traffic”, well-intentioned or not, that is no longer in touch with reality. Many studies have shown that on Twitter, 5% of user accounts are fake and 20% of traffic is generated by robots. Fine analyses carried out in these conditions often lead to confusion. We can also ask, considering the rapid development of socialbots as well as their sophistication, if this will corrupt social networks and cause their usefulness to deteriorate in the near future [RTS 16].
A socialbot [TEC 17] is a piece of software (a robot) that controls a social media account. This is a computerized system that will respond according to the type of network, but also according to the typology of user responses (for example on Twitter, it is easy to produce responses, as expression using 280 characters remains basic). Compared to a regular bot, the socialbot will share its responses by posing as a real user, which gives it more persuasion power.
There is controversy over socialbots since they may not be ethical; they are not real people even though social networks are designed for people. Since the socialbot tends to pass for a human being, they are only one step away from using identity theft to appear even more convincing. This is one of the main problems with socialbots, as they could, considering the importance of social network users, influence events by carrying out coordinated attacks on the best nodes (or influential individuals) sharing links.
We have therefore written this chapter considering this context, revealing the levers, restrictions and facilities but also threats generated by these systems.
We will present the different types of network, followed by some significant work carried out in the domain of “manipulating” and analyzing these networks, showing the positive and negative points.
6.1. Different types of social networks
There are hundreds of social networks. Some are used more in Europe, in the United States or Asia. Exchanges on these networks range from simple tweets to the exchange of emails, photos, videos, the sharing of documents, opinions, etc. Relatively closed communities can form or simple exchanges of information can take place.
It is however essential that even those who do not take part in these exchanges are aware of the number of individuals who do and of the increasing number of ever more varied networks accessible for free in most cases. Finally, social networks have a memory, often a very long memory, and what is shared on them will follow you, even several years later.
There are many articles discussing the main social networks [KAL 17, MAI 16, MEH 15, MIL 15, MOR 17]. Here is a typology of accessible networks:
| Typology | Examples of networks |
| Social Media Sites | Facebook, QQ, Skype, Twitter |
| Photo-sharing Networks | Snapfish, Flickr, Pinterest |
| Lifestyle Networks | Last.fun, Cross.tv, Car2 |
| Travel Networks | Couchsurfing, TravBuddy |
| Mobile Networks | Cellfun, Itsmy |
| Video Networks | YouTube, Vine, Youku |
| Reunion Networks | Classmate, Mylife |
| Business Networks | LinkedIn, Viadeo, Xing |
| Pre-teen and Young Adult Networks | Weeworld, Twenti |
| Blogging Networks | Xanga, Plurk, Livejournal |
| International Social Networks | Mixi, Netlog, Renren |
These networks have an impressive number of active accounts and to give a simple idea of the number of users, we refer the reader to the table on this subject indicated on Wikipedia [WIK 18].
In this table, we see that some sites have nearly two billion registered users and more than one billion active accounts. The list here is far from being complete and one of the things a business should consider when using these networks is warning employees about the dangers inherent in posting documents relating to their jobs or to the business. They should not attempt to control which networks are used by its employees, since this is next to impossible.
Figure 6.1. List of virtual communities with more than 100 million users [WIK 19]. For a color version of this figure, see www.iste.co.uk/dou/strategic2.zip
6.2. General remarks on social networks
One important thing to bear permanently in mind is that social networks give a false impression of security and respect for privacy. On the one hand, each document you place on these networks remains undestroyed, even if you press the “delete” button. On the other hand, all social networks have administrators (sysadmin) who can permanently see what is posted and exchanged. Do not think that this only happens in the United States. There may be private companies or mafia groups doing this. It is therefore necessary to be very careful when you “post” documents or exchange messages or answer questions, even from your friends, since as we have seen above, their identity may be stolen.
Although these recommendations are good for businesses, they are also valuable for individuals, all the more so as from answer to answer and question to question, you can easily be led to say certain things, to reveal certain documents that could harm you not only in the moment, but for years to come. And do not forget either never to use computers and networks accessible from your work station for private activities even if you have permission to do so under some conditions. Also, pay attention to facilities for accessing networks offered to you restaurants, hotels, airports and stations; do not say anything via messaging that you would wish to confine to a private conversation. Similarly, do not post photographs that could turn out to be compromising. If you have to send documents, they should often be anonymized as far as names and numbers such as pricing suggestions, quotations, etc. are concerned.
6.2.1. Why use social networks in a business?
The development of company social networks [ENL 13] has happened over the past 10 years. The development of these networks was motivated mainly by the fact that human resources saw them as a means of creating cohesion in the business, a feeling of freedom and well-being favorable to its smooth running. Then, “marketers” saw the opportunity via these same networks to promote certain products, make the arguments needed to promote them part of a mindset so that promotion moves outward from the business’ circle of employees:
“Internal networks (SelectMinds, Leverage Software, etc.) reserved for employees began to appear in large businesses such as Dassault and France Télécom.”
Various discussion take place on these networks, some about various activities, but others more centered on problems linked to the life of the business. We thus see the appearance in these networks of individuals who have the most connections, those who seem to be leaders in this space and who are not necessarily those who direct the business or are responsible on the ground. It is in this way that a sort of semi-virtual dualism is created between the actual business and the virtual business present in the network. In this space, it should not be thought that all is for the best, as frustrations and criticisms can be expressed, decompartmentalizing problems and making them visible to the outside world. This will give rise to an interpenetration between the business’ “public sphere” and the private sphere, which will not occur without creating problems and can lead to the business losing its image in the outside world. The question of interpenetration between public networks (such as Facebook) and private networks arises, which, again, will pose problems, namely of information being leaked or the private network being penetrated by third parties. It will therefore be necessary to put in place a security policy, but also an information policy for personnel, in order to avoid this [PER 14].
6.2.2. The risks of social networks in a business
It is necessary for a business to protect itself from the risks inherent to social networks. Not only should staff be aware of general risks, but it is also necessary to put in place a policy on use or a charter of good conduct for employees regarding networks. The risks run are relatively well-known:
- – the business’ e-reputation (it is necessary to check regularly what is being said about the business on networks), not only in blogs or discussion forums, but also by watching videos that might involve it. We only have to think of videos circulating about abattoirs to see the “damage” they may do to a business’ image. There are a great many instances of “harmful” videos, videos in fact posted to cause harm, or posted “for a laugh” without considering the harm they can cause; the leaking of information, ranging from hacking, to voluntary or involuntary leaks, to copying by third parties (photos of prototypes), expertise, etc. It should not be forgotten that fake profiles can be created and it must be ensured that a technical request made by a colleague really does come from them (one should telephone them before answering);
- – cybercrime in general. Fake requests coming from banks, updating one’s bank details, phishing, comments from fake profiles, etc. In the context of cybercrime, we distinguish two main levels [AUS 05, CAS 17]: Type 1 cybercrime: this is an occasional event, such as placing a Trojan (a type of malicious software) on your computer that installs a piece of software that will follow the strokes on your keyboard, this can happen for example by clicking on a link sent to you in an e-mail. This also involves security failures in a software making it possible to Type 2 cybercrime: this involves continuous interaction with the target. Making contact on a discussion forum for example install viruses, Trojans, “rootkits”, etc., for purposes such as bank fraud, etc.; then little by little, creating a link with the person contacted to carry out a wrongful act (sending money or secret technical information, etc.). Also in this category are actions carried out on forums by terrorist groups, etc.
6.3. The dangers of social networks
One of the most significant dangers in the domain of social networks, whether they are public or private, is linked to “guided socialbots” [ELY 16]: socialbots that will be directed to a specific network in an organization to penetrate it and obtain information (for example by creating fictitious identities or by stealing existing identities). The development of these guided socialbots also results from the already indicated fact that users of social networks are rarely aware of the total loss of privacy on these networks [FIR 14, SOU 17]. Because of this, users share personal data or information linked to the life of their business which is often sensitive and can lead to a considerable increase in security risks from attacks on the network, guided socialbots, identity theft, sexual threats, etc. [DES 14, LIN 09, WOL 10]. We say that these aspects arise mainly from personal security, but the security of the business may also be broadly involved. Similarly, the spreading of rumors or negative comments can create certain damage without detecting the source of these rumors or comments. Finally, beyond the business, when social networks involve state organizations, such as the army or political parties for example, this can lead to the development of threats affecting the security of the country, mainly with the aid of collecting targeted information and the development of rumors.
What is very interesting in the work of Aviad Elyashar [ELY 16], is that he has shown using two social networks (Facebook and Xing: “as we learned from executing our suggested algorithm most of our socialbots were able to infiltrate specific employees on both Facebook and Xing”) that it was perfectly possible to infiltrate specific employees, who tend always to accept and answer requests from strangers as well as from friends. The authors have also shown that in the space of five to six days, they were able to contact more than fifty users using socialbots, which are very easy to create. When accepting a friend request, it is therefore important to know what the potential risks are and to pay attention to the questions asked and the answers provided. For example, out of a total of 219 friend requests sent to 219 users of a network X, 88 users accepted the request and a 131 rejected it. These 219 users moreover included 10 targeted users (targeted for level of competence, access to information, ability to answer easily, etc.) and of these, four answered positively, which is a total of 40%. This illustrates the risks well and poses a question linked to the users of business social networks: should work experience students have access?
6.4. Minimizing negative influence on social networks
We have seen that social networks are a favorable place for developing real information, but also false information. Although the majority of authors are interested in the development of influence on networks (whether positive or negative), some studies currently focus on minimizing negative influence involving an individual, institution, business, political party, etc. Different authors have shown that malicious information is generally spread more quickly than positive information [BAU 01, CHE 10] and that by analyzing receiver profiles, an individual tends to more firmly believe information coming from several of their “friends”. It is therefore fundamental, in a social network, to decide the nodes (individuals for example) that will exercise the greatest influence in order, in the process of minimization, to block them using one of the three following processes [CHA 16]:
- – an approach based on blocking nodes: a node is represented by an individual who spreads (information in the present case). Blocking this node will lead to a drop in the frequency with which it is spread;
- – an approach based on blocking links: a link is formed by the transmission of information from one node to another. The problem is recognizing the minimal number of links needed to be effective. Blocking links will generally lead to a drop in diffusion frequency;
- – an approach based on competitive influence [KAU 15]: the nodes most active in diffusion are marked, and diffusion of counter-propaganda or publicity, etc. from these nodes is implemented.
In the work presented by Zakia Challal [CHA 16], more technical data are presented; they demonstrate that this direction is feasible at IT level. They therefore merit being studied in more detail. The example developed by Chen Wei et al. is interesting, as it shows how influence or counter-influence actions can be carried out without illegally penetrating the network’s general graph:
“[S]ocial networks are owned by third parties like Twitter, LinkedIn, Facebook, etc. The proprietary of the social graph is kept secret for privacy as well as company benefits. The owner of the social network is called the “host” and companies trying to run the viral campaigns are called the “clients” for the hosts. Clients cannot access the social network directly and hence they cannot choose seeds for their campaign by themselves. Clients would need the host’s permission and privilege to run. Motivated by this observation, we propose and study the naive problem of competitive viral marketing from the host’s perspective. In this study, we consider a business model with the host offering viral marketing as a paid service to its clients. The clients will hence be able to run the campaigns by specifying the seed budget, i.e. the number of seeds desired. The host of the social network controls the seed selection and allocation to companies.” [KAU 15]
In this example, the authors consider two types of approach, one where the nodes alone are considered and the other where the edges (links between nodes) have positive or negative influence.
6.5. An example of an international social network: the Confucius Institute
This example makes the link between soft power [NYE 04] or influence, which was developed in detail in a previous chapter, and the use of networks and communication facilities, making is possible to unite communities and spread various information on an international scale in quasi-real time. Thus, China [MIK 15], in order to develop influence operations at global level, developed the Confucius Institutes [KOU 17] as well as Chinese lessons provided by Chinese educational institutions.
6.5.1. Public diplomacy and Confucius Institutes
This creation of influence at international level between the domain of public diplomacy is used by China; the Confucius Institutes form part of this [LU 15]. Generally, we distinguish two aspects in this diplomatic approach, one is classical, which is the diffusion of targeted information to a very broad public: this is the top down system; the other, which is more “modern” (new public diplomacy) relies mainly on developing networks [HOC 05]. More precisely, public diplomacy mainly involves five elements [CUL 08, FLE 04]:
- – listening: collecting information on international opinion, especially using secret methods, such as espionage and information gathering;
- – advocacy: promoting policies, ideas or specific interests to foreign publics, generally via a country’s own ambassadors in other countries;
- – cultural diplomacy: promoting a country’s cultural resources abroad and/or facilitating cultural transmission abroad;
- – diplomatic exchange: promoting the reciprocal exchange of individuals between nations, e.g. reciprocal student exchanges;
- – international broadcasting: the use of radio, television, broadcasting and internet communication to engage with foreign publics.
In the context of networks, we are particularly interested in the last three points.
6.5.2. Structuring the network of Confucius Institutes
If we look globally at the structure of the network, it is clear that the Confucius Institutes have grown very quickly [HAR 14]. The number of institutes as well the number of Chinese classes was distributed across the world in 2013 according to the following table [HAR 14].
Table 6.2. Global distribution of Confucius Institutes and Confucius classes
| Regions | Confucius Institutes | Classes |
| Europe | 149 | 153 |
| The Americas | 144 | 384 |
| Asia | 93 | 50 |
| Africa | 37 | 10 |
| Oceania | 17 | 49 |
These bodies are all linked to Hanban University [HAN 14] which forms the center of the network (the headquarters). After that, the network becomes more complicated since all Confucius Institutes are linked to one another, including both the institutes themselves and also the Chinese classes linked to partner Chinese universities. Finally, the network’s complexity extends fully with the participation in the network of individuals present or engaged in the Confucius Institutes. Thus, the whole creates a relational structure through which a dynamic will emerge [ZAH 14] which is key to developing the Confucius Institute network. The aim is, particularly via online activities, to stimulate activities that will enable sustainable collaborations. Understanding this relational dynamic is a key point. In many cases and among others with the system developed by the USA we are dealing with a system of mass distribution, which is based solely on the receptivity of individuals but not on their interactivity. It is therefore necessary for activity promoters to model and control their messages constantly to maintain a certain initiative. In the case of the Chinese approach, the situation is different. We adopt a network communication system that will create a dynamic, making it possible to develop from the three main dimensions:
- – the structure of the network that facilitates contacts and information exchanges;
- – the synergy that will develop from the previous exchanges;
- – the co-creation developed by members of the network especially by presenting experience, teaching methods, results obtained, etc. This co-creation can of course go further.
This dynamic thus transforms users who are passive at the start into “stakeholders” and thus dynamic actors, which means that the network’s traction develops by itself from the exchanges, co-creation and narrative stories developed by members of the network. It is in this way that through its culture, China is developing a soft power differential [WAN 10]. Therefore, proof of the network’s vitality lies not only in the number of Confucius Institutes, but the links these various institutes develop, whether it is in the “Chinese general headquarters”, (the University of Hanban), or links developed between institutes within a single country or between countries. The social network of Confucius Institutes can be visualized as follows:
- – first layer: all the Confucius Institutes in universities across the whole world, Chinese partner universities who act as hosts and the online portal for the Confucius Institutes are all linked to the University of Hanban (the headquarters);
- – the second layer is formed of the link between the Confucius Institutes in foreign host universities and Chinese partner universities. The network is becoming multidirectional;
- – the third layer is formed of the link between the Confucius Institutes in foreign host universities in the region;
- – the fourth layer is formed of the link between a Chinese partner university and multiple foreign universities hosting Confucius Institutes;
- – the fifth layer is formed of the link between Confucius Institutes and foreign host universities plus the link with Chinese partner universities.
If we add to this (institutional) network the different links that may be created (during meetings, conferences, involving the financing of institutes, projects, etc.) between different individuals physically belonging to the network, considerable robustness is achieved within the system. It is this whole that makes it possible to co-create projects, promote initiatives and maintain the network’s growth via exogenous development. We therefore understand how the online part takes on considerable importance since it decreases distances, “shortens” timespans, facilitates contacts and creates greater cohesion.
6.6. Examples of software enabling analysis of social networks
We present two types of analyses. One, the analysis of Twitter, is specific to a network while the other involves knowledge of a tweet propagating for example in social networks. So, we have one specific approach and one more general approach.
6.6.1. Analyzing tweets
We can currently access archives [CHA 14, CUS 13] of our tweets within the parameters of our Twitter account by clicking on the “request your archive” tab and access them as a file. From this, different analyses can be made. Here, we present an extract of a tutorial created in the framework of the Pegasus project that, in a very detailed article, explains various results as well as ways of obtaining them [ROC 13]. The following sample is used by the author:
“The sample is formed of 17,744 tweets (as of March 1st, 2013, 00:15). Distributed over a period of 1,683 days, this makes on average around 10.5 tweets per day.”
In Figures 12.2, 12.3 and 12.4 we show some results of analyses obtained by the author. These analyses, if they represent all an individual’s tweets, are also important, as they allow you to know what a third party can learn from analyzing your tweets.
Figure 6.3. Word cloud representation of hashtags used [ROC 13]
6.6.2. Sentiment mining or opinion mining
“Accounting analysis” of tweets or emails is helpful, but it has a limit, as it does not consider “sentiments” present in the content. It is in this context that sentiment analysis was developed.
This analysis is based mainly on three aspects: positive, negative and neutral (i.e. positive mixed with negative). To carry out these analyses, we generally have recourse to dictionaries of words or expressions, verbs and adjectives. Depending on the number of times these terms occur in a text (or in a set of phrases contained in a text), we can give this text or these phrases a score.
But this technique has a limit, as the text needs to be processed before analysis, to be “lemmatized”, that is for example all verbs need to be put into the infinitive, nouns into the singular, etc. then we must consider negations, etc. Generally, to do this, we use dictionaries which are created both manually and automatically using usual terms, expressions, etc.
This is difficult, on the one hand because the language (French, English etc.) must be considered, but also because in social networks spelling mistakes, abbreviations, acronyms (for example “lol” = laughing out loud) often appear. Here are some examples of dictionaries: General Inquirer [GEN 17], WordNet-Affect [STR 04], SentiWordNet [ESU 06], all three English-language and ANT USD [WAN 16] for Chinese.
In their book Opinion Mining and Sentiment Analysis, Dominique Boullier and Audrey Lohard [BOU 12a] present the theoretical and practical aspects of these different types of analyses. In particular, they underline the role of experts in the domain, as a purely automated processing could contain multiple errors or may not be sufficiently refined [BOU 12b].
In this regard, we can mention:
“Sentiment analysis (sometimes called opinion mining) is the part of text mining that tries to define opinions, sentiments and attitudes present in a text or set of texts. Essentially developed since the 2000s, it is used especially in marketing to analyze for example commentaries made by internet users or evaluations and tests by bloggers or even social networks: much of the literature on this subject involves tweets, for example. But it can also be used to test public opinion on a subject, to seek to characterize social relationships in forums or even to check if Wikipedia is really a neutral medium.
Sentiment analysis requires much more understanding of the language than text analysis and classification by subject. In fact, although the simplest algorithms consider only the statistical frequencies with which words appear, this generally proves to be insufficient to determine dominant opinion in a document, especially when the content is short, such as messages in a forum or tweets.” [EDU 15]
On a practical level, there are various offers on the market, making it possible to class data as positive, negative or neutral, mostly involving analysis of tweets.
6.6.3. A more general approach: analyzing tweets in social networks
Here, we show a result obtained using Talkwalker [RIC 16]. In a campaign linked to #BacktotheFuture, Nike decided to create self-adhesive boots and send them to Michael J. Fox. It does not need stating that this initiative created a frenzy among Internet users.
In Figure 6.5, we can therefore show a virality map of Talkwalker.
Figure 6.5. Representation of the way in which the initial tweet (on the left) has “travelled” on social media [RIC 16]. For a color version of this figure, see www.iste.co.uk/dou/strategic2.zip
We can thus, from this map, analyze the overall reach of the initial tweet.
This is detailed in Figure 6.6 which shows how a Japanese journal contributed to diffusing the initial tweet in Japan.
Figure 6.6. Information concerning the initial tweet and its redistribution [RIC 16]. For a color version of this figure, see www.iste.co.uk/dou/strategic2.zip
6.7. Beyond socialbots and other IT systems, human action: fake news
We have seen the influence of computerized systems such as socialbots, identity theft, etc. These activities, which are fact initiated without continuous human action, are important and should be considered. But, on social networks and considering the importance of users, “classic” human action can be developed without restraint. It is in this way that information that is true but also information that is false will circulate on social networks: this is fake news. This false information can be placed on networks either intentionally for political or commercial ends or indeed simply as a game by real users. As the majority of this information will not directly “impact” an individual or business, the authors of this fake information cannot be prosecuted and so will be able to continue with their actions. Even if counter-information campaigns are instigated, it has been proven that even so, 30% of individuals who receive this fake information will remain receptive to it.
6.7.1. The fake news dynamic
In an article published in Futuribles [SOU 17], Walter Quattrociocchi [QUA 17a] suggests three factors to explain the dynamic of fake news. First, comes functional illiteracy, that is the inability to understand a text properly. According to the OECD (Organization for Economic Cooperation and Development), this affects many individuals between the ages of 16 and 65 in France and Italy. Then, the cognitive bias known as “confirmation bias” plays a major role, as each person tends to favor information that confirms their opinions or worldview. Finally, the Internet itself should be questioned since content is sent and received without an intermediary, such that no authority controls the veracity or basis of what is put online. This situation is particularly damaging, as there are no “regulators” on social networks and so no active controls on the information made available to users. We therefore return to a sort of educational vaccination that should serve as an antidote to fake news. But this is affected by users’ educational levels and especially their ability to practice critical thinking; this is explained in Chapter 2 of this book. It is in this way that very strange ideas have been permitted by a proportion of users such as: “the Americans never went to the moon, the whole thing was filmed in a studio”, “September 11th was caused by a succession of explosions inside the Twin Towers in New York”, “Europe is the cause of redundancies taking place in French businesses”, “spinach provides the body with more iron than any other food”, etc. We must however emphasize recent action such as that taken by France Culture in its own broadcasts [BRO 13, BRO 17]. You can access a podcast of these interviews and take a more critical look at information recounted on this network.
6.7.2. Beyond publishing online
Fake news and rumors pose a problem that goes beyond publishing information online:
- – it is the speed with which it spreads, and so its retransmission that, in a way, give it the appearance of truth: “everyone is talking about it”. It is in this way that people influence one another;
- – the use of data on social networks produces statistics that appear to validate information. However, after analysis, this information really includes a great deal of false information, thus falsifying results that were previously considered true.
In this way, fake news will influence individual opinions and significantly so, since for example in politics, many election results are separated by only a few percentage points. Since everyone can express themselves freely on social networks, this phenomenon is far from diminishing, but will be amplified by different actors, whether in politics, economics or in mafia activity. Walter Quattrociocchi has therefore indicated:
“For these reasons, we are witnessing a real mass phenomenon involving misinformation (partial or garbled information) or disinformation. Moreover, in 2013, the world economic forum, an independent international foundation that debates the most urgent problems on the planet, cited the mass diffusion of fake information as one of the gravest threats our society is facing.” [QUA 17b]
The Internet therefore acts as a resonator and will amplify rumors not only when they propagate but also, and this is more serious, by reinforcing those aspects of them that are true. Trials have been undertaken to provide some decoding keys making it possible to verify conspiracy theories circulating on social networks. The French newspaper Le Monde [DEC 17] for example presents a step consisting of verifying the following points:
“One group is always pulling the strings; detail is presented as absolute proof; coincidences that become proof; the absence of a reliable source becomes an additional argument; conspiratorial rhetoric does not allow questioning: sometimes it is not possible to explain everything in the wake of an event; take care not to see conspirators everywhere.”
We are not really in the domain of actual lies, but the presentation of real events from a particular angle. So, the best barrier against accepting false news still remains education, as well as the practical use of each person’s ability to know and analyze this information to put it into critical perspective, either alone or even better, as part of a group.
6.8. You love, you “like”, you click, you evaluate, but beware of “click farms”
Since social networks were created, sharing, “liking” to show your agreement with a tweet, choosing particular applications depending on the number of clicks they generate, has been a reality. This system has been imposed and many individuals are receptive to it, apply reason and make choices according to it. An application for a smartphone for example will be chosen from a list considering its position, generated in fact by the number of clicks it has, etc. This practice has spread and is becoming a commercial challenge. We are seeing the emergence of a new practice, a real economic war spreading across all countries. This re-thinking by Antonio Casilli [CAS 10] has given rise to a debate showing how fake news and click farms responsible for disinformation are coming to form a global market [ROO 17]. From its origins in propaganda, “word-of-mouth”, myths and plots, we now have “alternative facts”, the use of bots and a vast market of likes and fake clicks.
6.8.1. Calling Facebook into question?
As Casilli underlines, the functioning of Facebook may pose a question. For example:
“The restriction of the organic reach of messages on Facebook is a process initiated by the platform to prompt its users to move towards a model where they pay to share their posts. Publications that are not sponsored are naturally limited to your own community: to have a more satisfactory reach rate, one must spend money […] Political parties understand this well.”
6.8.2. Click farms
Click farms were created in this context. These are varied organizations that, for a fee, will increase the number of clicks from “followers”, etc. Recently, several of these have been discovered by the police. Tens of thousands of telephones, or indeed more, are available on racks linked to computers that will automatically generate clicks. Moreover, to make this activity more realistic, hundreds of thousands of SIM cards are used to ensure those generating clicks are rotated. It was in this way that, according to the Bangkok Post, the Thai authorities discovered 474 telephones and 347,000 SIM cards, all destined to interact with the Chinese network WeChat [MIC 17]. In Russia, it is possible to buy clicks in supermarkets. One can therefore, for a few rubles buy “likes” from “followers”. This practice was filmed by a Russian journalist [BIG 17] in the Okhotny Ryad shopping center in the center of Moscow. One can boost one’s popularity on Instagram or Vkontakte (a Russian social network similar to Facebook) in this way and for 50 rubles (about €0.78 or US$0.89) you can have 100 likes on a photo or for 100 rubles (about €1.57 or US$1.78) a hundred “followers”.
Information on these click farms is becoming more and more specific, for example a Russian journalist [EXC 17] shared a video on zerohedge.com that shows how a Chinese click farm was run and, according to the author, this click farm was formed of around 10,000 telephones. Other videos can also be accessed and show how this activity is organized, making it possible to add likes, followers and posts [20M 17a]. This practice, although it is not well-known, has existed since 2014; Zero Hedge revealed at this time that million followers on Twitter cost US$ 600 [DUR 17] and that the US State department had bought 2 million Facebook fans. In 2017, BBC News [BBC 17, VOL 17], revealed the discovery of 350,000 fake Twitter accounts. Research carried out by different groups of researchers, especially at the University of South Carolina and Indiana University [VAR 17] showed that between 9 and 15% of Twitter account users are bots [NEW 17]. It is therefore clear that even though some socialbots are beneficial [FER 16] such as those spreading of news or scientific publications, the appearance of new, unethical practices poses a major problem. For example, Donald Trump’s Twitter account has 31 million followers, but it is estimated that of this number, half are fake [BIG 17]. A response may be organized around campaigns to delete fake accounts or to complain about individuals spreading false reviews in the case of Amazon, for example [20M 17b].
6.8.3. A new type of fake news
The recent development of voice morphing as well as the manipulation of video sequences of public figures are leading to the creation of very high-quality fake videos. In a short time, considering technical advances, specialists will be able to modify a video, not to modify the image, but to modify the words uttered by the person:
“The combination of voice-morphing technology with face-morphing technology will create convincing fake statements by public figures. Given the erosion of trust in the media and the rampant spread of hoaxes via social media, it will become even more important for news organizations to scrutinize content that looks and sounds like the real deal.” [SOL 17]
6.9. Big Data
Many books and scientific publications have appeared in this domain and it is difficult, in a general presentation, to analyze them in depth [ERL 16]. Only to clarify, we will say that the production of information linked to queries in various search engines available on the market, e-mails, social networks, the online market via the Internet, the use of various applications on smartphones and now the ever-more significant emergence of connected objects, is becoming so substantial that it is impossible for the human brain to analyze it. All this information is generally known as the sixth continent that, virtually, gathers together a body of more or less interconnected data. From this immense and ever-growing stock of information, ever more numerous applications are being developed to analyze the information contained in this body statistically [DOU 14].
From these analyses arose the concepts of smart cities, co-operative healthcare systems (sharing data to obtain personalized care), managing electricity distribution via smart grids, water management, etc. Applicable to all domains, these masses of information, subject to ever-more effective analyses, are both an aspiration and a threat. In fact, this data may involve fairly technical domains (transport, the smart grid, etc.) which in itself do not pose a problem. However, it can interfere with our daily life since either without our knowledge or with our consent, we provide vast quantities of data about our behavior, which makes it possible to profile either groups of individuals or specifically us personally as consumers, but also as users of varied networks or as searchers on the Internet for example. It is in this way that more and more often, algorithms will “type” our behavior. At the limit, if this were only for commercial operations for instance, this would be a lesser evil. But the system is more malign, as it will tend to “push” the information we wish to hear or see toward us. A self-certainty is forming around many people that will smother any critical thinking since the only information provided us is information that fits our “view of things”. Although it has been noted that many people spend more time or at least as much time in front of their computer screens or mobile phones than in front of the television, we see that a single view of events and a single analysis of them is leading to self-deception, which will influence our behavior in various situations such as leadership [MEH 17], or decision-making for example [DIN 17]. This self-intoxication is often involuntary, albeit linked to one’s e-environment, and will often lead to extremes due to a lack of any critical sense [SMI 17]. But this may go further and become a sort of conditioning for human beings. For example:
“[…] it will not stop there. Some software platforms are moving towards ‘persuasive computing’. In the future, by using sophisticated manipulation technologies, these platforms will be able to guide our actions, whether to run complex work processes or to generate free content for Internet platforms, from which businesses gain billions, or to cause ‘political’ developments, etc. The trend is shifting from programming computers to programming people.” [HEL 17]
We therefore find ourselves at a crossroads where we can move either toward a digital democracy or can return to a form of feudal system if evermore powerful algorithms remain in the hands of only a few deciders. Every possible ethical question is therefore being asked about the protection of private information, etc. a widely debated topic whether in the domain of healthcare [BER 16] or in other domains [PUC 16].
6.9.1. The development of Big Data analytics
Big Data analytics now has fantastic research potential, but also potential for revenue. It is counted in billions of dollars and the impact of computer processing now accounts for 1.5% of global electricity consumption. We could continue in this vein with hundreds of figures on the production of emails, the exchange of SMSs, transactions on the net, etc. [RAI 16]. But that is not the question. Indeed, one of the problems posed by processing these gigantic data masses is the question of how pertinent the correlations that will be found are, and especially their relationship to one another [CAL 17]. Some authors maintain that “very large data bases are a major opportunity for science and data analysis is a new domain of investigation in IT. The effectiveness of analysis tools is used to support a philosophy that rejects scientific methods as they have developed through history. From this viewpoint, correlations discovered by computer should replace understanding and should guide prediction and action”. Consequently, it might not be necessary to give a scientific reason for phenomena observed, by suggesting say, causal relationships, since patterns from very large databases are enough: “with enough data, the figures speak for themselves.” But, the same author shows that “very large databases must contain arbitrary correlations. These only appear due to the size of the database and are not due to the nature of the data. They can be generated at random, in databases of sufficient size, which means that most correlations are false. Too much information tends to behave in the same way as very little information. Scientific methods can be enriched by ‘Big Data analytics’ used in immense databases, but this should not replace them”.
This therefore means that we should be circumspect and that the role of data scientists should be complemented by individuals able to provide real meaning for the correlations obtained, through analysis and reflection. But for the citizen, it is also necessary for the sake of transparency, that the results of this processing can understood, intelligibly, by non-specialists and this is true for very varied sectors, among others in the domain of mapping [GAU 16].
6.10. Conclusion
Beyond the diverse analyses that are possible and will certainly be refined more and more, a general remark should be made on these social results. These are not certain, the main consequence is the fact that before spreading information of whatever kind: personal, work-related, photos, videos, comments, etc. we should ask ourselves: is this information likely to harm me, either now or later? Similarly, can this information harm third parties, can it lead to the disclosure of critical information, etc. This is good practice and should always be present. This is especially true when using Twitter where very often, tweets are made or answered reactively, that is while leaving our emotions to guide our answers. This can be very dangerous and you could present a potentially harmful image of yourself. For example, we can cite the series of tweets coming from diverse individuals after the 2016 election of the president of the United States, Donald Trump. In fact, the initial tweets made after the election results were known were often replaced by better thought-through tweets some hours or days later.
A second observation should be made: on social networks, a large proportion of traffic is created by robots, and it is possible to create fictitious identities using socialbots to create a buzz or even to steal identities in order to ask questions or spread noise. If these aspects develop, we might fear the worst: the overall image taken from examining social networks would be largely false and so taking strategic decisions based on this would lead to the worst results.
We should always bear in mind that human tendency is to consider news sent by one’s friends as true, or indeed by several of one’s friends at the same time, as true. We should therefore remain vigilant when we use these media. They are certainly very useful, but in some cases, precaution must be taken, especially when handling confidential data, or responding on controversial topics. Finally, some networks, especially public networks, are not as safe as private networks, although these are still capable of being penetrated by socialbots.
Finally, there are now commercial services that analyze social media for you. It is certainly interesting to use these, but the results should be subject to critical analysis and evaluated before making important decisions based on the results.
6.11. References
[20M 17a] 20 MINUTES, “Bienvenue dans une ‘ferme à clics’”, May 14, 2017, available at: http://www.20minutes.fr/high-tech/2067747-20170514-video-bienvenue-ferme-clics.
[20M 17b] 20 MINUTES, “Amazon porte plainte contre 1 000 personnes accusées de publier de faux avis”, October 19, 2017, available at: http://www.20minutes.fr/web/1712867-20151019-amazon-porte-plainte-contre-1000-personnes-accusees-publier-faux-avis.
[AUS 05] AUSTRALIAN INSTITUTE OF CRIMINOLOGY, “High tech crime tools”, High Tech Crime Brief, no. 12, 2005, available at: https://aic.gov.au/publications/htcb/htcb013
[BAU 01] BAUMEISTER R.-F., BRATSLAVSKY E., FINKENAUER C. et al., “Bad is stronger than good”, Review of General Psychology, vol. 5, no. 4, pp. 323–370, 2001, available at: http://www.seekingbalance.com.au/wp-content/uploads/2016/06/BadStrongerThanGood.pdf.
[BBC 17] BBC NEWS, “Massive networks of false accounts found on Twitter”, January 12, 2017, available at: http://www.bbc.com/news/technology-38724082.
[BER 16] BÉRANGER J., “La valeur éthique des Big Data en santé”, Les Cahiers du numérique, vol. 12, pp. 109–132, 2016.
[BIG 17] BIG BROWSER, “Comment les Russes peuvent acheter des “likes” et des “followers” en supermarché”, Le Monde, June 13, 2017, available at: http://www.lemonde.fr/big-browser/article/2017/06/13/comment-les-russes-achetent-des-likes-et-des-followers-en-supermarche_5143705_4832693.html#wYWW5ZSTSCAYUMZa.99.
[BOU 12a] BOULLIER D., LOHARD A., Opinion mining et Sentiment analysis: Méthodes et outils, OpenEdition Press, Louvain-la-Neuve, 2012, available at: http://books.openedition.org/oep/214.
[BOU 12b] BOULLIER D., LOHARD A., “Opinion mining. État de l’art et exemples d’application”, Lingway, 2012, available at: https://www.slideshare.net/amandinet/opinion-mining-etat-de-lart-et-exemples-dapplications.
[BRO 13] BRONNER G., La démocratie des crédules, Presses universitaires de France, Paris, 2013.
[BRO 17] BRONNER G., KLEIN E., “Ce qu’on sait, ce qu’on ne sait pas : du vrai et du faux sur Internet”, France Culture, October 28, 2017, available at: https://www.franceculture.fr/emissions/la-conversation-scientifique/ce-quon-sait-ce-quon-ne-sait-pas-15-du-vrai-et-du-faux-sur-internet.
[CAL 17] CALUDE C.-S., LONGO G., “The deluge of spurious correlations in Big Data”, Foundations of Science, vol. 22, no. 3, pp. 595–612, 2017.
[CAS 10] CASILLI A.-A., Les liaisons numériques : vers une nouvelle sociabilité ?, Le Seuil, Paris, 2010.
[CAS 18] CASE.LU., “Cybercriminalité”, 2018, available at: https://www.cases.lu/cybercriminalite.html.
[CHA 14] CHARNAY A., “Vous pouvez désormais retrouver tous vos tweets depuis 2006”, 01Net.com, 2014, available at: http://www.01net.com/astuces/vous-pouvez-desormais-retrouver-tous-vos-tweets-depuis-2006-632867.html.
[CHA 16] CHALALL Z., BOUKHALFA K., “Minimisation de l’influence négative dans les réseaux sociaux. État de l’art et ouvertures de recherche”, Congrès Inforsid – 34th edition, Grenoble, France, May 31–June 3, 2016, available at: https://pdfs.semanticscholar.org/1ef5/5e41b95697fc7c57468c5118fb8fbf716a4b.pdf.
[CHE 10] CHEN W., YUAN Y., ZHANG L., “Scalable influence maximization in social networks under the linear threshold model”, IEEE International Conference on DataMining, Sydney, Australia, January 20, 2010, available at: http://ieeexplore.ieee.org/document/5693962/.
[CUL 08] CULL N., “Public diplomacy: taxonomies and histories”, The Annals of the American Academy of Political and Social Science, vol. 616, no. 1, pp. 31–54, 2008.
[CUS 13] CUSTINDA, “Comment récupérer l’archive de tous ses tweets ?”, TWOG, October 2, 2013, available at: http://twog.fr/comment-recuperer-larchive-de-tous-ses-tweets/.
[DEC 17] LES DÉCODEURS, “Décodex : comment reconnaître une théorie complotiste ?”, Le Monde, January 23, 2017, available at: http://www.lemonde.fr/les-decodeurs/article/2017/01/23/decodex-comment-reconnaitre-une-theorie-complotiste_5067727_4355770.html.
[DIG 17] DIGMANDARIN, “Confucius Institutes around the world – 2018”, 2018, available at: http://www.digmandarin.com/confucius-institutes-around-the-world.html.
[DIN 17] DINGS R., “Social strategies in self-deception”, New Ideas in Psychology, vol. 47, pp. 16–23, 2017.
[DOU 14] DOU H., “A new way to understand the “force field analysis” from Big Data analytics may be the future engine of the smart cities development”, Beijing Symposium on Competitive Intelligence, Peking University, 2014, available at: http://docplayer.net/11964003-A-new-way-to-understand-the-force-field-analysis-from-big-data-analytics-may-be-the-future-engine-of-the-smart-cities-development.html.
[DUR 17] DURDEN T., “A Russian went inside a chinese click-farm: this is what he found”, Zerohedge, May 11, 2017, available at: http://www.zerohedge.com/news/2017-05-11/look-inside-chinese-click-farm-fake-followers-fake-likes-fake-reviews.
[EDU 15] EDUTECH WIKI, “Analyse de sentiment en text mining”, 2015, available at: http://edutechwiki.unige.ch/fr/Analyse_de_sentiments_en_text_mining.
[ELY 16] ELYASHER A., FIRE M., KAGAN D. et al., “Guided Socialbots: infiltrating the social networks of specific organizations’ employees”, AI Communications, vol. 29, pp. 87–106, 2016.
[ENL 13] ENLART S., CHARBONNIER O., “Réseaux sociaux : opportunité ou menace pour les entreprises?”, Économie Matin, August 8, 2013, available at: http://www.economiematin.fr/news-reseaux-sociaux-chiffres-entreprises-utilisation.
[ERL 16] ERL T., KHATTAK W., BUHLER P., Big Data Fundamentals: Concepts, Drivers & Techniques, Prentice Hall Press, Upper Saddle River, 2016.
[ESU 06] ESULI A., SEBASTIANI F., “SentiWordNet: a publicly available lexical resource for opinion mining”, Proceedings of the 5th Conference on Language Resources and Evaluation – LREC’06, Genoa, Italy, May 24–26, 2006, available at: http://nmis.isti.cnr.it/sebastiani/Publications/LREC06.pdf.
[EXC 17] EXCUISITE M., “Voici une ‘ferme à clics’ dans laquelle on élève du like et de l’avis positif certifiés 100 % non-bio”, Journal du geek, May 15, 2017, available at: http://www.journaldugeek.com/2017/05/15/voici-une-ferme-a-clics-dans-laquelle-on-eleve-du-like-et-de-lavis-positif-certifies-100-non-bio/.
[FER 16] FERRARA E., VAROL O., MENCZER F. et al., “Detection of promoted social media campaigns”, Proceedings of the 10th International AAAI Conference on Web and Social Media, Cologne, Germany, May 17–20, 2016, available at: https://www.aaai.org/ocs/index.php/ICWSM/ICWSM16/paper/view/13034/.
[FIR 14] FIRE M., GOLDSCHMIDT R., ELOVICI Y., “Online social networks: threats and solutions”, IEEE Communications Surveys & Tutorials, vol. 16, no. 4, pp. 2019–2036, 2014.
[FLE 04] FLEW T., “Confucius Institutes and the network communication approach to public diplomacy”, The IAFOR Journal of Asian Studies, vol. 1, no. 1, pp. 1–18, 2004.
[GAU 16] GAUTREAU P., NOUCHER M., “Information géographique numérique et justice spatiale : les promesses du ‘partage’”, Justice spatiale-spatial justice, no. 10, 2016.
[GEN 17] GENERAL INQUIRER, Website, 2017, available at: http://www.wjh.harvard.edu/~inquirer/.
[HAN 14] HANBAN, Website, 2014, available at: http://english.hanban.org/.
[HAR 14] HARTIG F., “The globalization of chinese soft power: Confucius Institutes in South Africa”, in R.-S ZAHARNA, J. HUBBERT, F. HARTIG, (eds), Confucius Institutes and the Globalization of China’s Soft Power, Figueroa Press, Los Angeles, pp. 47–66, 2014.
[HEL 17] HELBING D., FREY B.-S., GIGERENZER G. et al., “Will democracy survive Big Data and artificial intelligence”, Scientific American, February 25, 2017, available at: https://www.scientificamerican.com/article/will-democracy-survive-big-data-and-artificial-intelligence/.
[HOC 05] HOCKING B., “Rethinking the “new” public diplomacy”, in J. MELISSEN (ed.), The New Public Diplomacy: Soft Power in International Relations, Palgrave Macmillan, Basingstoke, 2005, available at: http://culturaldiplomacy.org/academy/pdf/research/books/soft_power/The_New_Public_Diplomacy.pdf.
[KAL 17] KALLAS P., “Top 15 most popular social networking sites and apps”, Dreamgrow, February 26, 2017, available at: https://www.dreamgrow.com/top-15-most-popular-social-networking-sites/.
[KAU 15] KAUR H., Blocking negative influential node set in social networks: from host perspective, PhD thesis, Kennesaw State University, 2015, available at: https://digitalcommons.kennesaw.edu/cgi/viewcontent.cgi?article=1673&context=etd.
[KOU 17] KOUMA J.C.G., L’Implantation des Instituts Confucius dans le monde : un vecteur de puissance pour la Chine ?, Thesis, Université de Yaounde II-Soa, 2017, available at: http://www.memoireonline.com/09/11/4824/mLe-facteur-culturel-dans-la-cooperation-sino-camerounaisele-cas-de-limplantation-de-linstitu12.html.
[LIN 09] LINDAMOOD J., HEATHERLY R., KANTARCIOGLOU M. et al., “Interring private information using social data”, Proceedings of the 18th International Conference on World Wide Web, Madrid, Spain, April 20–24, 2009, available at: https://dl.acm.org/citation.cfm?id=1526709&picked=prox.
[LU 15] LU S., Analysis of the network communication approach to public diplomacy: case study of the Confucius Institute of the Carleton University, Master’s thesis, University of Ottawa, 2015.
[MAI 16] MAINA A., “20 popular social media sites right now”, Small Business Trends, May 4, 2016, available at: https://smallbiztrends.com/2016/05/popular-social-media-sites.html.
[MEH 15] MEHRA G., “105 leading social networks worldwide”, Practical Ecommerce, April 14, 2015, available at: http://www.practicalecommerce.com/articles/86264-91-Leading-Social-Networks-Worldwide.
[MEH 17] MEHMOOD A., “Leadership and self-deception: getting out of the box”, South Asian Journal of Management, vol. 24, no. 2, p. 201, 2017.
[MIC 17] MICHELOTTI L., “Fermeture d’une gigantesque ‘ferme à clics’ en Thaïlande”, Le Monde, June 13, 2017, available at: http://www.lemonde.fr/pixels/article/2017/06/13/fermeture-d-une-gigantesque-ferme-a-clics-en-thailande_5143890_4408996.html.
[MIK 15] MIKHNEWICH S., “The sage helps the celestial empire: the Confucius Institute as an instrument of China’s soft power in Greater East Asia”, International Organisation Research Journal, vol. 10, pp. 61–85, 2015, available at: https://iorj.hse.ru/data/2015/12/07/1095012358/The%20Sage%20Helps%20the%20Celestial%20Empire.pdf, 2015.
[MIL 15] MILANOVIC R., “The world’s 21 most important social media sites and apps in 2015”, SocialMediaToday, April 13, 2015, available at: http://www.socialmediatoday.com/social-networks/2015-04-13/worlds-21-most-important-social-media-sites-and-apps-2015.
[MOR 17] MOREAU É., “The top 25 social networking sites people are using”, LifeWire, April 30, 2017, available at: https://www.lifewire.com/top-social-networking-sites-people-are-using-3486554.
[NEW 17] NEWBERG M., “As many as 48 million Twitter accounts aren’t people, says study”, CNBC, March 10, 2017, available at: http://www.cnbc.com/2017/03/10/nearly-48-million-twitter-accounts-could-be-bots-says-study.html.
[NYE 04] NYE J., Soft Power: the Means to Success in World Politics, PublicAffairs, New York, 2004.
[PAT 14] PATTON D.U., HONG J.S., RANNEY M. et al., “Social media as a vector for youth violence: a review of the literature”, Computers in Human Behavior, no. 35, pp. 548–553, 2014.
[PER 14] PERETTI J.-M., “Évaluer les risques liés aux réseaux sociaux d’entreprise”, Les Échos, September 17, 2014, available at: https://business.lesechos.fr/directions-ressources-humaines/management/communication-interne/0203763763030-evaluer-les-risques-lies-aux-reseaux-sociaux-d-entreprise-103335.php.
[PUC 16] PUCHERAL P., RALLET A., ROCHELANDET F. et al., “La Privacy by design : une fausse bonne solution aux problèmes de protection des données personnelles soulevés par l’Open data et les objets connectés ?”, Legicom, vol. 1, pp. 89–99, 2016.
[QUA 17a] QUATTROCIOCCHI W., “Désinformation sur les réseaux sociaux : ce que révèlent les statistiques”, Pour la science, no. 472, February 2017, available at: http://www.pourlascience.fr/ewb_pages/a/article-desinformation-sur-les-reseaux-sociaux-ce-que-revelent-les-statistiques-38074.php.
[QUA 17b] QUATTROCIOCCHI W., “Désinformation sur les réseaux sociaux : ce que révèlent les statistiques, Part. 1, les complotistes sont-ils crédules ?”, Pour la science, no. 472, February 2017, available at: https://cyrilc42blog.wordpress.com/2017/03/24/desinformation-sur-les-reseaux-sociaux-ce-que-revelent-les-statistiques-par-walter-quattrociocchi-part-1-les-complotistes-sont-ils-credules-2/.
[RAI 16] RAISSON V., 2038, Les futurs du monde, Robert Laffont, Paris, 2016.
[RIC 16] RICHARD, “La carte de viralité. Découvrez comment les publications deviennent virales”, Talkwalker, October 27, 2016, available at: https://www.talkwalker.com/fr/blog/la-carte-de-viralite-decouvrez-comment-les-publications-deviennent-virales.
[ROC 13] ROCHAT Y., “Tutoriel : exploiter ses données Twitter”, Pegasus Data Project, March 27, 2013, available at: https://pegasusdata.com/2013/03/27/tutoriel-exploiter-ses-donnees-twitter/.
[ROO 17] ROOS G., “Fake news et ‘travailleurs du clic’ : comment la désinformation est devenue un marché mondial”, franceinfo, February 10, 2017, available at: http://www.meta-media.fr/2017/02/10/fake-news-et-travailleurs-du-clic-comment-la-desinformation-est-devenue-un-marche-mondial.html.
[RTS 16] RTS INFO, “Les algorithmes de Facebook accusés de promouvoir des fausses nouvelles”, October 13, 2016, available at: https://www.rts.ch/info/sciences-tech/reperages-web/8087792-les-algorithmes-de-facebook-accuses-de-promouvoir-des-fausses-nouvelles.html.
[SMI 17] SMITH M.-K., TRIVERS R., VON HIPPEL W., “Self-deception facilitates interpersonal persuasion”, Journal of Economic Psychology, no. 63, pp. 93–101, 2017.
[SOL 17] SOLON O., “The future of fake news: don’t believe everything you read, see or hear”, The Guardian, July 26, 2017, available at: https://www.theguardian.com/technology/2017/jul/26/fake-news-obama-video-trump-face2face-doctored-content.
[SOU 17] SOUPIZET J.-F., “Désinformation, ‘fake news’ et réseaux sociaux”, Futuribles, March 9, 2017, available at: https://www.futuribles.com/fr/article/desinformation-fake-news-et-reseaux-sociaux/.
[STR 04] STRAPPARAVA C., VALITUTTI A., “WordNetAffect: an affective extension of WordNet”, Proceedings of the 4th International Conference on Language Resources and Evaluation (LREC 2004), Lisbon, Portugal, May 2004, available at: http://wndomains.fbk.eu/publications/lrec2004.pdf.
[TEC 17] TECHOPEDIA, “Socialbot”, 2017, available at: https://www.techopedia.com/definition/27811/socialbot.
[VAR 17] VAROL O., FERRARA E., DAVIS A.-C. et al., “Online human-bot interactions: detection, estimation, and characterization”, Cornell University Library, March 27, 2017, available at: https://arxiv.org/abs/1703.03107.
[VOL 17] VOLTIGEUR, “Dans l’enfer d’une ‘ferme à clics’ chinoise”, Les moutons enragés, May 17, 2017, available at: https://lesmoutonsenrages.fr/2017/05/17/dans-lenfer-dune-ferme-a-clics-chinoise/.
[WAN 10] WANG J., Soft Power in China: Public Diplomacy through Communication, Palgrave Macmillan, Basingstoke, 2010.
[WAN 16] WANG S.-M., KU L.-W., ANTUSD: A large chinese sentiment dictionary, pp. 2697–2702, Academia Sinica, 2016, available at: http://www.lrec-conf.org/proceedings/lrec2016/pdf/450_Paper.pdf.
[WIK 18] WIKIPEDIA, “List of virtual communities with more than 100 million active users”, 2018, available at: https://en.wikipedia.org/wiki/List_of_virtual_communities_with_more_than_100_million_active_users.
[WOL 10] WOLAK J., FINKELHOR D., MITCHELL K.-J. et al., “Online ‘Predators’ and their victims”, Psychology of Violence, no. 1, pp. 13–15, 2010.
[ZAH 14] ZAHARNA R.-S., “China’s Confucius Institutes: understanding the relational structure & relational dynamics of network collaboration”, in R.-S. ZAHARNA, J. HUBBERT, F. HARTIG (eds), Confucius Institutes and the Globalization of China’s Soft Power, Figueroa Press, Los Angeles, pp. 9–30, 2014.