Visual Exploration of Complex Networks

jweebo writes "Seed magazine has a story on complexity, and how it can be visually represented with fascinating results. From the article: 'Complexity is everywhere. It's a structural and organizational principle that reaches almost every field imaginable, from genetics and social networks to food webs and stock markets ...Collected here are a few of the many intriguing, and often beautiful, images that illustrate how the whole is more than the sum of its parts.'"

Comments

Let me guess...

[jd]

You've been reading "Fractal Geometry of Nature" by Benoit Mandelbrot. Very nice illustrations and the section on how fractals all started and another on fractal dimensions were good, but otherwise the book was far too vague and had few proofs. This demonstrates Heisenberg's Writing Principle, which states that you can either know bout a topic or write about it, but not at the same time.

Wow! This is Unix!

[Speare]

Wow, this is Unix! I know this!!
--Lex

visualizing large-scale information

[Nick Mitchell]

as far as I can tell, the article only gives one quantification of the scale these folks are dealing with: on the order of tens of thousands for that one case. is this what is considered large? the point of visualizations is to show patterns of nodes (and patterns of paths) in graphy structures. at some point, one runs into one or the other of various limits:

1. pixels on the display: 2 million or so.
2. insufficiency of the clustering algorithms: showing one pixel per node and random placement, or placement by DFS traversal? for trees, or for graphs where classification is the primary concern, then tree-map or "Csoft" views scale relatively well in this regard, but what about for more general problems?
3. implementations (or algorithms) that don't scale: e.g. graphviz uses n^2 (n=#nodes) space for its graph layout!

one must always think about the summarization criteria: what aren't you going to show? how will you indicate that summarization has occured? how do you denote drill-down capability? what will the form of drill-down be? what heuristics should you use to selectively deaggregate, in order to highlight potentially interesting subgraphs? for large-scale info, this is as important as what you will be showing, and how it will be shown! for our stuff, we have graphs with tens of millions of nodes.

The point of visualization

[espressojim]

The point of visualizing data is to learn something that you could not do with the raw data. In all of the cases shown in the article (yes, I acually read TFA), I didn't spot an example where it actually showed anything useful.

The first example with proteins: how similar are two proteins? If two shapes are similar (and please, how many proteins where being graphed there? One, two, five?), then you might be able to recognize it. If they are similar shapes, are they always presented in the same orientation in space? Does color have any meaning? Does this graph have any legend? If I gave someone who understood the graphs two proteins, what could he say besides "these are related" and "these are not related"? We already have wonderful programs to compare two proteins and say how similar they are two each other, along with being able to the estimate significance of the measurement.

I'm not sure that the other graphics look more informative. They are all pretty, but if they do not convey information (and not lose a large amount of relevant information), then they are just a nice way to generate patterns for some nerd's tie.

Here is it:

[megaditto]

Dr. Wolfram (of Mathematica) offers PDFs of his book for free here (or pay $60 for hardcopy):

http://www.wolframscience.com/thebook.html

I do suggest you at least glance over the first few chapters, look at the pictures.

Also note that the guy got his PhD in Physics at the age of 18 I believe.