In past posts, such as the OSINT case on Twitter, we briefly introduced ourselves to how our environment defines us and affects us. We were able to see that it is possible to find out who is part of our personal environment thanks to how we relate to our contacts in social networks. Now, could we go deeper into Social Network Analysis?
I'm sure that if we evaluate any group of friends, we could identify other interesting data. Who hasn't seen that within the same group there are subsets that get along better than others or people with closer relationships with others. Whether it is to analyze a group of friends, a community in Social Networks or a terrorist group, there are two fields that help us study them.
What is Social Network Analysis (SNA)?
Social network analysis (or SNA) is an area of research focused on the study of a group of entities. These entities can be people, organizations or users, through the interactions that exist between them. This differentiates it from other social or behavioral sciences, since what it analyzes is how a network of individuals relates to each other. Therefore, we are not only talking about a network on Facebook or LinkedIn, but also on a physical level.
This field of study has developed different analysis methodologies with which to evaluate the behaviour of the network. Thanks to this, it is possible to see which are the most important entities in a network, who has more influence or which communities exist in it. In addition, there are many tools in the market focused on helping in this type of analysis.
What is SOCMINT?
The word SOCMINT refers to Social Media Intelligence, an intelligence discipline that encompasses all methods and tools to obtain and analyze text, photos, videos or any other material exchanged between two entities. In other words, it is a discipline that teaches us how to analyze the interactions between the members that are part of the network we are analyzing.
In this sense, SOCMINT makes use of the SNA as another tool of analysis through the study of graphs as the one we could see above in which we see people (nodes) and relationships (edges).
The class of 3B
Let's say we're secondary school teachers and we have a class of 30 students. These students know each other because they have been in the same class for three years. We, on the other hand, are new to the school and want to know which groups are in the class. To do this, on the first day each student answers a simple question: are you a friend of student X? X being each one of the rest of their classmates. We give you three possible answers:
- Yeah, we're best friends.
- Yeah, we're friends.
- We're not friends.
First to arrive, first to analyze
The results are not long in coming, and we have the first two students, who give us their results: Maria del Carmen and Carmen.
In this case, the darker relationships correspond to best friend relationships, while the lighter ones are simply friends. You may have met at a playground game sometime, but that's not who you hang out with most. Instead, those partners with whom they don't relate don't appear connected.
As we receive the rest of the answers, we can already see how both students influence the people they know best in their class. As you can see below, we can see 3 different groups of nodes:
- Maria del Carmen's friends (in blue), formed by all the nodes linked to her (18 nodes in total)
- Carmen's friends (in yellow), composed of those nodes attached to her (8 in total)
- A third group formed by all those nodes that remain loose since they do not connect with any other (in total 7 nodes in gray)
Also, we found a possible fourth group of nodes. Those in greener shades, which connect with Maria del Carmen and Carmen, and which would be more important than the rest of the coloured nodes. The reason? Because they connect with two of our students.
Degree (Degree Centrality)
With this information we already started to know the network. However, it is clear that we only know Maria del Carmen's and Carmen's friends, but we do not know how the other students relate to each other. After receiving the answers of the rest of the students of 3ºB, this is the resulting network:
As you can see, in addition to the relationships, the nodes have also been colored. Unlike the first ones, which tell us the type of relationship that exists, the more relationships the nodes have (whatever type they are), the darker they are. This factor is called Degree, and refers to, as in this case, the most popular nodes in the network by the number of connections they have. In this case, the more connections, the higher the degree.
We can see that besides Carmen and Maria del Carmen, another popular student in the class is Josefa. However, analyzing the number of connections of each student makes it difficult to understand how they relate to each other. Even more so if we consider that the greater the weight of the relationship, the more friends the students will be.
Below we will look at other ANS measures that will allow us to better understand the network. These measures are visualized in the network through the color of the nodes (students). The relations maintain the same criteria (pay attention to the type of relation), and, to remember at all times which are the students with more friends (grade), we are going to maintain the size of the nodes.
Intermediation (Betweenness centrality)
Betweenness is a measure that quantifies the number of times a node acts as a bridge on the shortest path between two other different nodes. Therefore, the more paths the student appears, the more popular he or she will be in the class. In this case, it would differ from grade centrality in that betweenness does not measure the quantity of connections a node has, but the quality of these.
Although this graph does not seem to provide any new information, it actually helps us to better qualify the popularity of the students in the class. First of all, we can confirm that both Josefa and Maria del Carmen are the best connected nodes in the network, even more so than Carmen or Juan. Secondly, and as can be seen in the figure below comparing the grade algorithm with that of Betweenness, it can be seen that the nodes of Antonio, Manuel, Pablo, Diego, María Teresa and Ana María lose popularity due to the quality of their connections (they have more friends than best friends).
Just as you can find out which students are most popular, you can also find out which ones are least popular. To do this, you would use the eccentricity algorithm, The eccentricity of a node is just the opposite of intermediation: the longest path from itself to any other node in the network. Nodes with less eccentricity are more central (and therefore more popular) by this measure.
Clustering (Modularity class)
Do you remember those little groups that used to form in class? Another interesting approach of the SNA is with subgroups (communities) within a network. In our example, knowing which students relate most to each other can be helpful in understanding how communication flows in the classroom.
In this case, four different groups can be identified. From more to less dark:
- Group 1: made up of 7 nodes; Manuel, Antonio, Lucia, Martina, David and, further away but apparently within the group, Lola and Laura.
- Group 2: formed by 10 nodes; Paula, Pilar, Carmen, María Teresa, Hugo, Ana María, Alba, María, Pablo and Daniel.
- Group 3: formed by 6 nodes that seem to be more dispersed; José, José Antonio, Diego, Josefa, Javier and Francisco.
- Group 4: formed by 7 nodes María del Carmen, Isabel, Sofía, Julia, José Luis, Alejandro and Juan.
Influence of relationships
Using a heat map, and taking advantage of the fact that different degrees of friendship have been established, the level of influence of the most popular members of the 3B class can also be evaluated. We talk about our already known Maria del Carmen, Carmen, Josefa and Ana Maria. This is important since this influence is understood not to be direct, but can also affect other students even if their relationship with the most popular students is weaker or even non-existent.
Maria del Carmen
In the case of María del Carmen, we see that her influence is much stronger with Lucía, José, María, Juan and the other three popular students of the 3B class: Carmen, Josefa and Ana María. The rest of the users seem to have more or less the same relationship with her.
Knowing what the influence of Maria del Carmen is, she could be a leader within the class, since she mainly relates to the rest of the most popular students. Therefore, she could be useful in case she wants to spread a message only among those students, since she seems to have a transmitting role rather than a diffusing one, and therefore they are more likely to pay attention to everything she says.
In Carmen's case we do find three different groups of influence, having more influence on some than on others. What stands out is that Carmen seems to relate to more individuals in the network than Maria del Carmen and, in spite of this, the latter seems to continue to be more important within the network.
The three groups are:
- Group 1: in which would be María del Carmen, Manuel, Antonio, David, José Luis and Pilar; showing themselves as darker nodes.
- Group 2: in which we would find Maria, Hugo, Maria Teresa, Ana Maria, Lola, Jose, Laura, Josefa, Pablo, Francisco, Diego, Isabel, Juan, Julia, Martina and Lucia.
- Group 3: formed by Paula, Alba, Javier, Daniel, José Antonio, Sofía and Alejandro; being the ones with less influence and, therefore, the ones with the lightest colour.
Given Carmen's influence on the web, in this case we see a different role within the class from that of Maria del Carmen, as in this case Carmen does seem to relate to many more students in a significant way, so the messages she can spread in the class would spread much more quickly, i.e. she would act as a speaker or disseminator.
In Josefa's case, the first thing that stands out is that she has no influence on any of the other popular students in the 3B class, so she is an actor who acts as a ceiling. We see that his influence is much more localized, only on Maria, Hugo, Jose Antonio, Isabel, Antonio and Jose Luis. This makes Josefa a niche broadcaster/speaker, i.e. someone who is able to convey messages to a very specific group of students in the class.
Finally, the case of Ana María is very similar to that of Josefa, since she is also a niche broadcaster who only connects with María, Daniel, José Antonio, María del Carmen, Diego and Julia, to whom she could transmit a message in a very focused way.
However, unlike Josefa, Ana María does have some influence on María del Carmen, which indicates that she can be useful in getting messages across to the latter since, of all the popular students in the 3B class, she has the most influence on the most important member of the class.
What is the point of knowing all these relationships and dynamics?
Performing social network analysis as in this case can be very useful when studying a community. It can be useful to monitor it in an efficient way, to be able to influence it or even to understand how it works in order to know how to use it in case the group studied has a relationship with us. For example, to understand how the employees of our organization or that of our competitor relate to each other, as explained in our article on Competitive Intelligence.
For example, it could help to define how to get a message to a certain member of the group through others, as the channel could be more direct, or even to find out who could be the best sender to achieve our goal. In fact, if this example was transferred to a community in Social Networks with influencers and followers and we worked for a marketing consultant, knowing how we could select which influencer should publish the message depending on whether we want a more massive diffusion or reach a certain audience. It would only be necessary to know how the diffusion of the message evolves depending on the influence of the users
Similarly, if this were actually a terrorist group and we wanted to see how it works through Telegram, we could get to know how they communicate through the messaging app, and control which are the most important communication channels to have them monitored.