#social network analysis

[phdmama]

What I’m Doing With My Very Expensive Data Science Degree

[ailelie]

A series of research studies found that gun violence—just like an infectious disease—can be transmitted from person to person in social networks: i.e., exposure to gun violence not only can lead to a host of negative psychological and cognitive outcomes but also increases the risk of individuals becoming gunshot victims themselves.

Furthermore, individuals who associate with a greater number of gunshot victims are at an extremely elevated risk of being victims themselves. For example, a study of a high-crime Boston community found that being directly connected to a gunshot victim increased one’s own probability of being a victim by 25 percent. This exposure effect also extends to indirect associations—meaning you do not have to be directly connected to a victim to experience heightened risk of victimization. For example, one Chicago study found that each social tie closer to a gunshot victim increased one’s probability of being shot by 57 percent.

(Emphasis mine)

Source: https://cops.usdoj.gov/RIC/Publications/cops-w0859-pub.pdf

[petrasthings]

Hannibal - Social Network Analysis

Hello everyone! I am a fannibal and I also study sociology at uni and this semester I took an introduction to social network analysis course and we were supposed to collect some data and analyse them. I chose all seasons from Hannibal and analyzed the interactions in a scene between all the characters :D Yes, I had a lot of fun doing this. I hope you’re ready for some science and a very sexy graph.

SNA (google it if you want to know more detailed information about it) is a type of analysis where we can study nodes and edges. Node can be a firm, an animal, a human or a character in a tv series. Edges connect the nodes, e.g., how often firms make transactions with each other or interactions between characters in a scene. There are many ways how to analyse these matrices and I was doing just some simple analysing because first, the course was just an introduction, and second, it would take a lot of time to collect more detailed data. So despite knowing I will “lose” some information in the proces (e.g., how long characters interact on the scene) I still have a very good base for answering the question: “Who is the main character in Hannibal?”

I’ve read some very impressive meta about it or the analysis of the names in the appearance of Mads’ and Hugh’s names in the Hannibal opening titles (link). But here I present to you some hard data.

So the basic numbers tell us, that the strongest connection is between Will and Hannibal (they have 114 interactions in the whole series), then it’s Jack and Will with 95 interactions, Jack and Hannibal have 55 interactions. Will and Alana have 38, Hannibal and Alana have 28. Then there are numbers around 20-50 interactions that Jack, Will and the team sassy science have with each other. What’s also interesting is that Hannibal and Bedelia have 28 interactions (same as Hannibal and Alana). And I should also mention that I was only counting the “real” interactions, not the things Will imagined in his head. E.g., when Will is talking to Elliot at the end of Coquilles (S01E05) or that iconic scene at the beginning of Shiizakana (S02E09) with Will, Hannibal and Wendigo and that terrific monolog about love - I did not count these scenes.

SNA works with visualisation of the matrices and also with measures. I was analysing (using UCINET) centrality measures, cohesion measures and subgroups. Let’s start with centrality measures. Degree is the number of edges that each node has. In the table below you can see that Will has the most connections in the matrix so that means he is the most central person. Then it’s Hannibal, Jack and team sassy science with Alana. Next is betweenness and that is a more complicated measure and it’s the ratio of the shortest paths and again, Will is on the first place. It tells us that Will has the highest betweenness because he has a lot of paths/connections/edges that go to a lot of other people in the matrix. But this measure also takes in account people, who might not have many connections but are close (in the matrix) to the most center nodes. So that explain why on the fourth place we have Abigail, then Bedelia, Freddie and also Molly. Then we have closeness, that’s the sum of the length of the shortest paths between the nodes (that’s also why with this measure, the smaller the number the better) and again, it’s very obvious, that Will is the most central person. After Hannibal and Jack, we have Alana, which is understandable considering her role in the whole series. The same goes with Freddie's fifth place - as a journalist you need to have connection to a lot of people to have the possibility to get a lot of information. Then we have Chilton and Abigail. And at last we have eigenvector which tells us that the central node is not central on its own, but it is the one who also have central neighbors - and here we have a surprising information. Of course that Will is the first again but Hannibal is the third this time! That means he doesn’t have, unlike Jack, many connections to people who would also be central and that’s very understandable. As a serial killer and a cannibal you don’t want to have that many connections, and also if we take in consideration what kind of people Hannibal likes (only Will really) or want to have connection to, it makes sense.

So it’s very obvious that Will, Hannibal and Jack are the main characters. You can also see that their centrality measures are much bigger numbers (except closeness where it’s the other way round) unlike with other characters. The measures of these three characters are very clear whereas the other characters (as you can see) are more complicated to analyze. I choose just those first 7 rows because from there it’s a real mess tbh.

With cohesion measures I analyzed a lot of boring numbers and I think that the measures that can illustrate this matrix are: average distance between nodes (characters) is 2.5 and that the longest distance between nodes it 6 which means that to get to anyone in the matrix you have to get to them max. through 6 nodes/people that are in the matrix.

And the last part of my analysis are subgroups. In many matrices we can find them. A type of subgroup, where everyone has a connection to one another, is a clique. The biggest clique that can be found is (in this order):

Will Graham, Hannibal Lecter, Jack Crawford, Alana Boom, Dr. Chilton, Dr. Gideon, Freddie Lounds. I could list here all the other cliques that the program in which I was analyzing this matrix has found but it’s not that interesting imo. Other cliques, that were found, were always (except one) with Will Graham. Which also corresponds with his centrality measures.

And so the answer to the question: “Who is the main character in Hannibal?” is: “Will Graham”, my dear fannibals.

And here is the sexy graph that I made (using Gephi), the more red the edges are, the closer are the nodes to the center. And as you can see, the strongest connection is between Will and Hannibal <3

[datascience-from-the-trenches]

Is Twitter transitioning from a content platform to a behavioral platform?

Retweeting has increased nearly linearly over time, from around 20% of all Twitter content in January 2012 to around 53% by October 2018.

Source: https://www.forbes.com/sites/kalevleetaru/2019/03/04/visualizing-seven-years-of-twitters-evolution-2012-2018/

Visualizing Seven Years Of Twitter's Evolution: 2012-2018

Looking back over seven years of the Twitter 1% stream, what can we learn about how one of the world’s most influential social networks ha

[college-out-of-context]

We're all suffering, but we think everyone else is perfect.

Social Computing Professor

[datascience-from-the-trenches]

Data science from the trenches turned 5 today!

[phaya]

Ponencia «DataFake: Early detection system for viralization of disinformation» en VII Jornadas de Comunicación Digital

Ponencia «DataFake: Early detection system for viralization of disinformation» en VII Jornadas de Comunicación Digital

Presentación del Trabajo Fin de Grado realizado por David Marcos Martín en las VII Jornadas de Comunicación Digital sobre DataFake, una herramienta para detección temprana de desinformación.

View On WordPress

[phaya]

Understanding discussions of citizen science around Sustainable Development Goals in Twitter

Understanding discussions of citizen science around Sustainable Development Goals in Twitter

David Roldán-Álvarez, Fernando Martínez-Martínez, Estefanía Martín and Pablo A. Haya, in IEEE Access, 2021, 9, pp. 144106-144120 https://doi.org/10.1109/ACCESS.2021.3122086 Abstract Citizen science (CS) involves volunteers who participate in scientific research by collecting data or by addressing the needs of the project they are involved in. In the last years, there has been an increasing…

View On WordPress