Share 0
Like 0
neuroscientist behind the keyboard - 9 - graph analysis and science

neuroscientist behind the keyboard - 9 - graph analysis and science

Humans, especially scientists, are good at applying problem solving approaches, found to be useful in one domain, to different domains and seeing what happens. In this blog I look at how often different aspects of graph analysis are associated with different domains in science. I use google search result again as basis of analysis, so enjoy the read and diagrams, but don't bank on the results.
In a previous blog, I skimmed over some academic domains that had seen the application of graph analysis. I felt that if I would have skimmed wider back then I would have sounded even less coherent and left it for another day. Since today is as good as any other day, here it goes. In another previous blog I played with google as a way to determine popularity. I figures this might be a fun way at a quick look at the landscape of that the association is between different aspects of graph analysis and some academic areas.

finding keywords for graph/network analysis aspects

Euler and the more mathematically researchers have used graph analysis for combinatorial and permutation analysis with great success. A single vertexes impact has been looked at to better define the influence an part of a graph has on others. Navigating from one vertex to another has been looked at by Dijkstra and others, as this speeds up finding ideal solutions/accessing data in computer science. Shapes and the effect of graph alterations (via removal/additions of edges) is another aspect that can be readily explored with graph analysis. The keyword chunks I came up with for these four domains can be found in
the table just below this line.

As for academic domains I went with names only this time. They were: astronomy, biology, chemistry, computer science, ecology, geology, medicine, neuroscience, physics, sociology, statistics.

what came back?

The associations of different academic domains with aspects of graph analysis is not what I expected it to be. So of them make intuitively more sense then others.

For example, that navigation is not very popular with chemistry makes sense to me as the navigation across bond structures is not something that comes up often in chemistry lectures. Also that combinatorics is the least favoured in astronomy kind of works for me. When you have billions of stars and trillions of light years and only one Stephen Hawkings, you don’t want to spend all your research time looking for combinations of problems which are hard to explain.

I was surprised to see that chemistry and medicine share a strong interest in shape and impact. Maybe this could be due to an interest in robustness of reactions/interactions and the impact of effects of parts of molecules on synthesis/health.

In the less sense space for me was that navigation had such a big association with sociology. I thought impact would dominate sociology, I was thinking that social networks, tribal structures and all would make this a clear case. While puzzling this tells me that I should maybe put my nose into a sociology book sometime to bring my expectations back to where they should be.

As I am digging my way through Sporns latest book, I am curious to see how the aspect fractions will change in biology and neuroscience over the next years. Which means, I might do an update blog some time, that looks at change in these aspects over years.


All of that is nice and good but how does this help the thousands of budding graph analysis prodigies which are reading my blog? Back when I was an undergrad, a lot of people left academia to work for investment companies to do protein and gene inspired analysis there. That was back in the yearly 2000s (I feel so old...). Anyhow, with the recent rise in social media I feel that there might be another rush to industry, this time for social network analysis.

So to give a different look on how big social media is now when it comes to graph analysis I did one more search. I took a look at the same four aspects and looked how often they were associated with the three big players (facebook, twitter, google+) in social networking. To be fair I compared them to the average of the previous science searches and not the sum.

I think the two figures and the tabele show nicely that science has some punch left when it comes to graph analysis, and it's not all facebook and twitter these days.

in summary

Network analysis is everywhere in science, we just like different flavours of it, depending on the nature of the problem we are tackling. Facebook, twitter and google+ are cool datasets and allow us to play with networks we didn’t have detailed information about in the past, but don’t let that distract you from giving Euler, Dijkstra et al another look.
Share 0
Like 0

You are in