===

Yes! You've uncovered an important rule of social network analysis, which is that social network graph images are useful with low numbers but quickly become difficult to understand meaningfully when the number of nodes vastly increases. Good work.

-- Prof. Cook

=== 

Busy, busy, busy, I agree! The clustering does help to organize the graph, but detail is hard to see. What's the URL for this image (created by NodeXL, by the way)?

-- Prof. Cook

===

Check out this image... wow, how interesting it is! It's not just the node size, which you note, but also the opacity of ties. These ties aren't simply black; their darkness appears to be related to the strength of tie they depict. That's a smart move that appears to counteract the disadvantage of lines that cross.

-- Best, Prof. Cook

The Sigiyama rules:

1. Lines should be straight.

YES lines are straight and easily discernible. 2. Lines should be far apart from one another.

Yes for the most part lines are spaced well apart and can be distinguished from one another. (Some are closer and a little more difficult to separate 3. Lines should not cross or touch.

Yes this rule is essentially followed...the majority of lines do not touch or cross...I assume as the network gets bigger with increasing nodes and ties that is most difficult to "stick to the rule" 4. Lines should be easy to follow from one node to another.

Yes- the ties (lines) in this graph are easy to follow 5. Nodes that connect should be close.

Yes no connecting nodes are unusually far apart. 6. Nodes that are most central should be close to the center of the graph.

It appears that the two MOST IMPORTANT nodes (or ones with the greatest betweenness centrality) are both are near the center (dark blue). Although I am glad I am NOT trying to figure this out...I am still struggling with that one (but the tutorial video did help) 7. Nodes that are similar in some way should be placed nearby one another.

Yes- there seems to be a clear color theme and relationship among the nodes. (the key ion the article explained the value of each color)

Although I don't know the specifics of this graph or what type of relationship/network it is looking at (I didn't get that far into the article) I can assume some relationships and I think it is a good and useful representation of a network graph and would work well to demonstrate the network and betweenness connectivity.

Wow! I just got really wordy there...hope that makes SOME sense!

===

You definitely make sense (although be sure to proofread!) I think this is a really, really interesting image. The color coding serves the image well in displaying betweenness centrality, although I'm glad that the Wikipedia article in question also describes the image verbally.

- Prof. Cook

1. Lines should be straight.

yes lines are straight 2. Lines should be far apart from one another.
Yes it respect the Kozo Sugiyama's guidelines 3. Lines should not cross or touch.
Lines are crossing and touching one another 4. Lines should be easy to follow from one node to another.
no, only few of them are easy to follow. 5. Nodes that connect should be close.
no they are not close to one another 6. Nodes that are most central should be close to the center of the graph.
No, central nodes are all over the corners 7. Nodes that are similar in some way should be placed nearby one another.

Yes they are. Yellow squares are nearby to one another

From my view I believe this a very busy social network and this is the only way it should be even though it is kind of confusing visualization. Some of the central nodes a isolated with their ties in all over the corner of the graph, some other are easy to follow from one another. In over all, this graph does not follow the Kozo Sugiyama's guidelines.

http://www.getperception.com/blog/2011/08/measuring-who-and-how-we-influence-via-social-network-analysis/

1. Lines should be straight. Yes, lines are straight.

2. Lines should be far apart from one another. No, lines overlap.

3. Lines should not cross or touch. No, lines appear to overlap.

4. Lines should be easy to follow from one node to another. No, there is some confusion in the dense areas.

5. Nodes that connect should be close.  Not necessarily, nodes appear to be arbitrarily spaced whether they are connected or not.

6. Nodes that are most central should be close to the center of the graph. No.

7. Nodes that are similar in some way should be placed nearby one another. It is hard to tell from this picture.

I think that while this graph is colorful and eye catching, it could have been done in a much clearer way. This graph is confusing as many of the edges blur together.

====

Vela, I agree: it's certainly a stylish image, but it's not at all clear what it signifies. Must aesthetics and information be at cross-purposes? -- Prof. Cook

I found this very interesting.

1-Straight Lines- Not all of the lines are straight.

2-Lines are far apart- Not all of the lines are equally separated.

3-Lines should not cross or touch- These lines all cross each other due to the nature of the graph.

4-Lines should be easy to follow- The lines are fairly easy to follow.

5-Nodes that connect should be close- No

6-Most central nodes should be centered in the graph- The priority points are found on the outside of the graph and the lines connect in the middle.

7-Nodes that are similar should be near to each other- The skills are grouped together on one side of the sphere while the careers are on the other.

This is not a traditional network graph. However, because of the color, and the concept I was looking for, it makes sense. This is memorable and easy to read and follow. Specific skills are focused on one half of the sphere while careers are on the other. The lines connect to show what skill is useful for what career. While it doesn't follow a traditional networking graph, I feel it provides the same general information required - how is one thing connected to another.

===

Libra, I really like the systematic way you evaluated this Williams College network graph using Kozo Sugiyama's standards. I like the image, too, especially because if you click through and hover over the image, it's dynamic: the node you focus on is rendered visible and all connections NOT involving that node drop away.

-- Best, Prof. Cook

https://www.washingtonpost.com/news/worldviews/wp/2015/09/24/the-future-of-language/?tid=sm_fb

====

I like this image a lot. Sure, there are some overlaps, but the style of spoke-and-hub radiation from centers expresses a lot about the idea of centrality.

- Prof. Cook

It definitely does not follow Kozo Sugiyama's guidelines. Although there are some straight lines, there are also curved lines which are touching everywhere and not far apart. They are not easy to follow; in fact, it's too hard to tell if the nodes that are close are connected and vice versa. They do use different colors, shapes and sizes. I think that the visualization here is confusing and it reminds me of a giant ball of rubber bands. :)

===

Vega, you can say that again! I have to tell you, I have NO idea what this graphic means except for "Wow, there are a lot of nodes and ties here." In defense of the graphic, it is possible to click on the graphic and zoom in at the Sunlight Foundation blog, but even then it's not very legible. I agree with you!

-- Prof. Cook

Edit: Here's the link! https://threatpost.com/researchers-graph-social-networks-spot-spammers-061711/75346/

===

Hello, Phoenix! Thanks for uploading this image, which I agree is overwhelming. The color and shape of nodes might offer us a clue -- is there a clear explanation on the originating website of why some nodes have some shapes and colors? By the way, could you also include the URL (web address) where this image can be originally found on the web? Thank you.

-- Prof. Cook

====

Cassiopeia, thanks for sharing this image. I'm not sure exactly what you mean here. Some of your phrases ("once the information exceeds... the rules are not followed") don't have a clear meaning. In the last sentence you rush through the numbers of Sugiyama's rules without explaining yourself. Always do your best to explain what you mean. Provide examples! Is it really true that nodes that connect are not close? If so, could you describe a few examples of that problem?

- Best, Prof. Cook

1. Lines are straight.

2. Lines are (relatively) far apart.

3. Several of the lines do cross or touch it appears.

4. The lines appear easy to follow from one node to the next.

5. Connected nodes are close to each other.

6. The most central node seems to be in the center of the graph.

7. Similar nodes seem to be grouped together.

http://blog.crossjoin.co.uk/2010/02/22/nodexl-network-graphs-and-the-eurovision-song-contest/

====

Yowza -- this COULD be a very interesting graph. But I find I can't follow the patterns -- there are just too many lines seeming to head anywhere. I bet this graph could benefit from a little of that bundling function of NodeXL's that I show off in the videos. Are the nodes arranged by geography?

Best, Prof. Cook

http://blog.revolutionanalytics.com/2013/03/a-map-of-worldwide-email-traffic-created-with-r.html

In this network, email traffic throughout the world is analyzed.  The lines are in fact straight however they are not far apart from each other and the lines do cross.  With that being noted, the network is still viewable as the colors make it easy to see the broad depiction.  You can see that the blue and green (Hindu and Islamic), appear to have a high density.  They both are the only pathway to get through to others as well, so both would possess a higher betweenness centrality than others.  While you can see from one node to another the ties in the middle of the network are difficult to differentiate from one another.  Moreover, the placement of the nodes has been laid out properly in regards to the guidelines.  While the network could have a wider spread to make it more viewable, it seems to provide a good outlook on email traffic projections in the world. 

https://dataist.wordpress.com/2012/03/11/combining-d3-and-raphael-to-make-a-network-graph/

I have attached the link within the post due to a computer error on my older mac. I will try to post from a school computer next week to avoid this issue.

http://support.sas.com/documentation/cdl/en/grnvwug/62918/HTML/default/viewer.htm#p0q343kxjyj36jn1e2z6lulkda3j.htm

Multi-level force-directed: The  application combines graph partitioning and spring or force-directed  layout heuristics to determine node positions.

Lines are straight. The lines are far apart. There are lines that touch and cross. The lines are easy to follow from node to node. The connected nodes are close to each other. The most central node is in the center of the graph. Similar nodes are grouped together.

===========

Thanks for the update, Sagitta. I appreciate it. And I see: this network graph was made by the statistical package SAS to show that in SAS you can make a network graph! Aha! That makes it not surprising they'd highlight a clean, clear graph as an example.

- Best, Prof. Cook