Data

Different combinations of terms were established to develop a search strategy divided into six blocks to retrieve all the tweets in Spanish that highlight the main purposes of exploitation linked to human trafficking. These combinations are composed of different root terms that refer to trafficking in persons per se and are complemented by specific words for each type of exploitation: a) sexual; b) labour; c) for begging; d) for forced marriage; e) for the commission of crimes; and f) for forced removal of organs.

Accordingly, each tweet was labelled considering the type of exploitation to which it referred, and some of them could appear in more than one. It is assumed that by naming one or more of the word combinations included in Table 1MC (supplementary materials), a tweet is introducing one or more types of exploitation into the activity and public debate on this phenomenon. In addition, other data were extracted from the tweet, such as the actor who posted it and the mentions made.

The terms associated with each of the exploitation purposes were selected according to different inclusion criteria. The first criterion was the relevance of the terms, defined as their potential to locate a large number of tweets for a particular purpose and their ability to represent some of the most important concepts associated with that purpose. Secondly, the relevance and unambiguity of the terms were assessed, defined as the degree of specificity of the term for the type of exploitation it addresses and the number of potentially misleading results its use could lead to, either because it does not refer to the exploitation purpose in question or because it includes results referring to other purposes.

The download of tweets was carried out on 8 June 2021 using Graphext’s Tractor ¹ tool, setting 31 December 2020 as the deadline for the publication of tweets and retrieving tweets, retweets and replies. However, tweets dated before 2011 were removed from the analysis due to the low activity prior to that year (two crime tweets, 84 labour tweets, seven marriage tweets, three begging tweets, 18 organ tweets and 764 sex tweets). The final set of tweets consists of 101,051 tweets, of which 90,328 are unique.

Method

Firstly, a descriptive analysis of the tweets was carried out using R. Descriptive statistics were applied and the activity was explored in general and for each type of exploitation in particular, at the level of tweets and actors and differentiating between active actors (those who tweet about trafficking) and mentioned actors. Semantic networks were also generated with Graphext based on the text of the tweets. In these networks, each node represents a tweet and the connections between pairs of tweets are given by their semantic relationships. By detecting communities using the Louvain algorithm (Blondel et al., 2008), the main topics of interest were identified and the main communities were tagged after manually reviewing the tweets and the most relevant terms in each of them. Two researchers were involved in this labelling task, reaching a consensus on the solution in the event of discrepancies. In both analyses, the activity was considered for the full period (2011-2020) and also for two different periods (2011-2015 and 2016-2020).

Evolution of general activity for each exploitation type

Of the 101,051 tweets analysed (90,328 unique tweets), the most frequent exploitation purpose linked to human trafficking is sexual exploitation (Table 1). Not only does it account for the largest volume of tweets (66,615), it also has the highest number of unique tweets, making it the purpose most frequently addressed exclusively (60,301 tweets; 90.52%). It is followed by labour exploitation (13,493 tweets; 69.2%), while the rest are much further behind. The type of exploitation with the lowest number of tweets is ‘for commission of crimes’ (242 tweets; 54.88%).

If we take sexual exploitation (the purpose most frequently present among the tweets) as a reference to analyse the overlap between tweets that talk about two exploitation types, it converges with 25.54% of the tweets that address labour exploitation, 31.89% of ‘begging’, 32.59% of ‘forced marriage’, 31.97% of ‘commission of crimes’ and 23.66% of ‘forced removal of organs’. Thus, sexual exploitation is not only the exploitation type most frequently represented exclusively, it also comprises roughly a quarter of the tweets about other exploitation types, sharing some of the attention they receive.

The evolution of the level of presence of the different exploitation types in tweets over time has a series of common and divergent features. Firstly, an analysis of the distribution of tweets based on a density graph (Figure 1) shows that activity on Twitter began timidly in 2010 for most exploitation types. Some events in 2010 that could be related to the beginning of the Twitter debate activity include: a) the fact that Twitter use became popular at that time; b) consolidation of international guidelines and national legislation, especially relating to the codification of the crime of human trafficking (General Congress of the United Mexican States, 2012; United Nations, 2000).

The exponential increase in activity is concentrated between 2017 and 2018 in the case of sexual and labour exploitation, begging, forced marriage and commission of crimes and during 2016 and 2017 in the case of forced removal of organs. Most countries that have ratified the Palermo Protocol (2000) had criminalised both human trafficking as a stand-alone offence and any kind of purpose of human trafficking by 2016 (Sweileh, 2018). In the case of exploitation for forced organ removal, the growth in visibility from 2017 onwards could be related to the dissemination of news based on UNICEF reports pointing to the high percentage of people exploited for this purpose in Mexico (Arena Pública, 2017). In fact, Mexico’s National Human Rights Commission (CNDH) implemented a strategy in 2019 that aims to collect statistical data at the state level on people affected by this phenomenon, among others, with the involvement of the State Attorney-General’s Offices (FGRs) (CNDH, 2019), which are also very active in promoting prevention campaigns that have an impact on this greater visibility.

Following this exponential growth during 2017 and 2018, the cumulative activity on exploitation for begging, for the commission of crimes, and to a lesser extent, for forced removal of organs, declined. Exploitation for marriage and especially labour and sexual exploitation did not experience such a notable decrease, with the activity even increasing for the latter two.

It is worth noting that sexual and labour exploitation have the highest volume of tweets, so the variability of their growth in the previous graphs is not as dependent on such a small number of tweets as might be the case for other types of exploitation, suggesting consolidation of the debate in general terms.Some of these fluctuations in activity may be related to the presence of certain peaks of activity on Twitter (Figure 2), identified based on days or periods with a higher number of tweets, although the events to which these peaks relate vary according to the exploitation type and do not represent the total amount of activity for each exploitation type. A large proportion of the tweet production peaks is related to the celebration of different international days sponsored by various international actors and organisations. Among these, World Day against Trafficking in Persons is particularly noteworthy, it ihas been organised since 2013 by the United Nations. This event seems to stand out, in particular since 2018, although there are other prior activities related to this Day, and it also seems to be related to peaks in all exploitation types. The International Day for the Abolition of Slavery, held since 1985 and also sponsored by the United Nations, is more prominent in the case of exploitation for marriage, and in 2020 for sexual and labour exploitation. In contrast, the International Day against Sexual Exploitation and Trafficking of Women, sponsored by the World Conference of the Coalition Against Trafficking in Persons and the Women’s Conference, seems to be linked to the main peaks of sexual exploitation, and, although to a lesser extent, also to publications on other types of exploitation such as labour exploitation.

Also, noteworthy are the media impact of the dismantling of a network of human traffickers who engage in forced begging and, even more so, the implementation of informative campaigns on human trafficking carried out for days or weeks by the FGRs of different Mexican regions. These usually refer to several exploitation types and, on fewer occasions, to just one, as in 2017 with the campaigns relating to exploitation for forced removal of organs destined for trafficking. Other events that also seem to lead to increased activity in the debate are the celebration of the World Day against Child Slavery and International Women’s Day; the increase in agricultural activity, given that a high percentage of trafficked persons end up working in this sector (UNODC, 2018; 2020); the implementation of campaigns in Spain —#trabajoforzoso— in 2019 and 2020; and the migration crises relating to the migrant caravans in Mexico (05/11/2018 and 18/01/2019), where migration seems to be linked to trafficking and organised crime and where institutions such as the Mexican regional FGRs seem to highlight exploitation for organ trafficking (second caravan).

Analysis of the actors participating in publications and their activity

The number of active actors —those who tweet— who are mentioned and the average dissemination of their activity in terms of tweets, retweets, favourites, and replies also varies depending on the exploitation type and the period: 2011-2015 or 2016-2020 (Table 2).

Although sexual and labour exploitation have the highest volume of active actors during both periods, the number of actors tweeting increased in 2016-2020 for all exploitation types. Most actors are mentioned in relation to sexual exploitation, especially in the period from 2016-2020. This suggests a greater inquiry to other actors in the tweets posted, recognising them as relevant in the debate. Other exploitation types such as the forced removal of organs also have a higher average number of mentions than the rest.

On the other hand, there is also an increase in the dissemination indicators referring to the number and average number of tweets and the average number of retweets, favourites and replies for all exploitation types during the period from 2016-2020. However, while the average number of favourites is higher in the case of labour exploitation than sexual exploitation, there is a greater increase in the number of replies in the case of sexual exploitation. This might suggest that while actors are directly aligned with other tweets on labour exploitation without introducing their convergent or divergent opinion, activity on sexual exploitation seems to elicit greater participation and expression of actors’ own opinions on third-party posts.

The emphasis on sexual and labour exploitation is also reflected in the number of actors talking about one or several exploitation types (Figure 3), considering that at least 20% of the tweets made by an actor must be about an exploitation type to be assigned to that group. The highest volume of actors corresponds to those who talk only about sexual exploitation (28,707), followed by labour exploitation (4,310), those who talk about both sexual and labour exploitation (3,344) and those who talk about forced organ removal (1,324 actors).

The actors posting the highest volume of tweets for each exploitation type vary for the periods between 2011-2015 and 2016-2020 (Table 2MC). During the period from 2011-2015, the activity of actors such as social organisations, academics and activists stands out, as is the case for sexual and labour exploitation. On the other hand, especially in the case of exploitation for forced organ removal but also exploitation for forced marriage, actors related to the criminal prosecution of human trafficking stand out, such as the police, the Ministry of the Interior and an international network to counter organised crime. During the period from 2016 to 2020, which had a substantial increase in activity, actors related to the criminal prosecution of human trafficking generally gained relevance. The presence of FGRs, Public Ministries, Ministries of the Interior, their advisors and police forces (all from Latin American countries) stands out. Although less frequently represented, the media and some social organisations are also present. In the case of sexual exploitation, for example, the most active actor was a social organisation.

The most frequently mentioned actors also vary according to the exploitation type and period (Table 3MC). During the period from 2011-2015, some of the most frequently mentioned actors are the media and audiovisual products (e.g. films), anti-trafficking campaigns, initiatives or institutional bodies, politicians and some social organisations. International organisations such as UNODC and UNICEF also appear, albeit discreetly. It is precisely during the period between 2016-2020 that these international organisations become particularly relevant. For almost all exploitation types, the UNODC division in Mexico, the UNODC itself as an umbrella organisation, and the United Nations stand out. Other bodies involved in the criminal prosecution of human trafficking such as the Guardia Civil and the FGRs are also noteworthy. In addition, personalities such as Pope Francis, López Obrador (President of Mexico), the documentary filmmaker Mabel Lozano, and institutions such as the CNDH stand out.

Identifying topic communities in the tweet network

The main communities of topics identified in the general network of tweets on trafficking and exploitation purposes during the period from 2011-2020 and the sub-periods of 2011-2015 and 2016-2020 (Figure 4) show that sexual exploitation is the purpose with the highest representation among the topics addressed.

The network for the entire period (2011-2020) includes a total of 90,328 nodes, each representing one tweet. 112 communities were detected, of which the 20 main ones were analysed (77.74% of the nodes are included in them). This network shows that the community with the highest volume of tweets (8% of the total) compares human trafficking with other crimes, making references to South, Central and North American countries. The communities relating to sexual exploitation, prostitution and pornography (7%), and prostitution (6%) are contiguous within the network. The International Day against Sexual Exploitation and Trafficking of Women is also well represented (7%, 3%, 3% and 2%). In addition, there are different communities on the criminal prosecution of the crime of human trafficking (4%, 3% and 2%); cases and convictions for sexual exploitation and prostitution (4% and 3%); social action against the crime of trafficking (3% and 2%), and legislation to combat this problem (1% and 1%).

An analysis of the evolution of the network between the periods 2011-2015 (25,515 nodes and 23 communities) and 2016-2020 (64,813 nodes and 84 communities), with 10 communities identified in each (94.99% and 62.6% of nodes, respectively), shows that most of them deal with sexual exploitation and prostitution. During 2011-2015 they represent the two largest communities (25% and 22%, respectively), although during 2016-2020, when the network fragments, these topics are presented transversally within the rest of the communities (Table 4MC). Prostitution, which seems to gain relevance, has an exclusive community (10%). Also noteworthy is the increase in references to criminal prosecution of the crime of human trafficking, different cases and their reporting, representing three of the 10 communities from 2016-2020 (21% of the nodes). During 2011-2015 there was also a large community on labour exploitation (16%), which decreased from 2016-2020 (6%), although two communities appear on other exploitation purposes and other crimes (6% and 6%).

The topic networks of the tweets were also analysed (2011-2020) for the two exploitation purposes most frequently present in the analysis: sexual (Figure 1MC) and labour (Figure 2MC). As it is the main purpose, the top 20 communities in the sexual exploitation network were identified. As can be seen, it is very similar to the general network of tweets where all exploitation types are included. On the other hand, and as in this general network, most of the communities in the network of tweets on labour exploitation, with 10 communities identified, also include references to sexual exploitation (Table 4MC).

The network on sexual exploitation focuses on topics such as prostitution and sexual exploitation (communities with 11% and 7% of tweets), together with references to pornography and women; other organised crime offences (7%) where the word ‘prostitution’ stands out, forming a community located in the centre of the network and close to the two previous ones, as well as that on sexual exploitation as a crime, especially child sexual exploitation (6%); criminal prosecution of sexual exploitation and human trafficking, reports of offences, convictions, and dismantling of networks (5%, 4%, 4%, 3% and 2%), some of which are located in the centre of the network; and the International Day against Sexual Exploitation and Trafficking of Women (5%, 3% and 3%).

In the network on labour exploitation, the majority community on the criminal prosecution of the crime of human trafficking and the institutions and agencies involved stands out (9% of tweets). Also prominent are other communities on labour exploitation which highlight its forced nature or slavery (6% and 8%); the victims of this exploitation (the previous community with 8% and another with 8%); different purposes of exploitation (7% and 5%); human trafficking in America (4%); and, finally, the quantification of people experiencing this phenomenon (4%), forming a larger community than in the general tweet network, although it is impossible to clarify whether it refers to victims of human trafficking, labour exploitation or both.