Interact & Explore the Merit Network

[Piggy]

This has been quite a long and challenging project given the amount of information to show, i have started to look into it since i saw the network graph threads of DdmrDdmr and sncc. My goal was to have an explorable version of the full network with interactions included, as well as my take on trying to find new merit sources.

Each node represent a user and the edges with arrows merit transactions, received or given.
I made 2 versions, the main difference is in the shape. Then for each of these versions i made two variants,one where the coloring is based on the Rank and the other is based on the Trust.








The node size is based on the Node Merit Size variable (on the User Information Panel) and reaches the maximum with a total of 500 merits given away, i did this to try to avoid the current merit sources shading emerging one. This should help visualize at sight the main givers just by looking at the size of the nodes.

Node Merit Size is calculated in the following way:

Code:

 If Merit Given > 500 
		Node Merit Size = 500
	   otherwise
		Node Merit Size = Merit Given

The node and edge color simply tells the rank/trust of the node.

I removed from the network all the nodes and edges where:

Code:

Merit Received + Merit Given = 1

So anybody who sent a single merit or received a single one won’t be in it.

There are exactly 15954 Nodes and 54108 Edges in there. If there are more merit transactions from user A to user B they will be grouped together and be represented with 1 line(Edge) that goes from A to B

What is the purpose?

You can use these visualizations to help identify users which could be a merit source, since at first glance you have a very intuitive way of knowing the amount of Merit given (size of the node) and Rank/Trust (color of the node). Plus the personal network tells the merit story of the user.

How i made this?

I prepared all the data i needed and made several iteration with Gephi for the layout of the network graph. The drawing is made using the library Sigma Js which supports WebGL.

How to use it?

• Each node you see can show the user name on mouse over and by clicking on it it will isolate the connections based on merit sent and received, both in the form of a graph and with a panel, which shows supplementary informations about the use selected as well. Moving the mouse over the names on the list will show the label on the graph as well.
  
• You can drag the network by clicking on an empty space, zoom in and out with your mouse wheel or using the buttons in the bottom overlay. The magnifying lens with the square center the graph on screen.
  
• On the left panel you can search for any user by name.

Links to the visualizations:

https://albertoit.github.io/Merit-Network-Visualization/
https://albertoit.github.io/Merit-Network-Visualization/?config=config_Trust.json
https://albertoit.github.io/Merit-Network-Visualization/?config=configsparse.json
https://albertoit.github.io/Merit-Network-Visualization/?config=configsparse_Trust.json

[1715145197]

1715145197

[1715145197]

1715145197

[1715145197]

1715145197

[1715145197]

1715145197

[TheBeardedBaby]

Great work as usual.
My small cloud is beautiful

[DdmrDdmr]

As I tend to do, I debugged the graph for my username and I saw two Donators on it. One is Ognasty, who appears with a decent sized sphere (the largest, along with other usernames) even though Ognasty has historically sent me 1 sMerit and I have not sMerited this account. The size is the same as other contributors who have awarded me way more merits, so I’m not sure the sizing is working 100% correctly.

It is indeed quite a lot of work to get there. It seems that you can get something similar on to Tableau, and I looked ad doing it on my Dashboard, but having to go through Gephi to derive the coordinates is not a quick process, and seems way too slow to allow me to do it weekly without too much additional overhead. I don’t know if you are planning frequent update of your tool, but I believe is a heavy task.

Nicely done and appreciate the effort behind the scenes.

Edit: I'm also getting a graph on the Trust graphs, but I'm not in the trust.txt file. Am I missing something ?

[yuzefa]

Really nice tools, this could be usefull for mods and admins.
I'm still thinking we need really much more Merit sources, and this could help to identify them faster.

 

[Piggy]

As I tend to do, I debugged the graph for my username and I saw two Donators on it. One is Ognasty, who appears with a decent sized sphere (the largest, along with other usernames) even though Ognasty has historically sent me 1 sMerit and I have not sMerited this account. The size is the same as other contributors who have awarded me way more merits, so I’m not sure the sizing is working 100% correctly.

It is indeed quite a lot of work to get there. It seems that you can get something similar on to Tableau, and I looked ad doing it on my Dashboard, but having to go through Gephi to derive the coordinates is not a quick process, and seems way too slow to allow me to do it weekly without too much additional overhead. I don’t know if you are planning frequent update of your tool, but I believe is a heavy task.

Nicely done and appreciate the effort behind the scenes.

Edit: I'm also getting a graph on the Trust graphs, but I'm not in the trust.txt file. Am I missing something ?

I capped on purpose the size of each node base on a scale from 1 to 500 merits sent (Node Merit Size field in the User Information Panel), otherwise we would be looking at enormous nodes of the merit sources which sent more than 1k merits, taking all the scene, this way you can notice also those users who sent perhaps "just" few hundred merits and are not merit sources.

In this case Gephi produces a gfex which i can then use directly to show the graph, you need anyway to prepare and clean the data and then i had to intervene directly on the file to add other information in the way i wanted (Trust, Rank and the scale for the node i mentioned above), it takes time anyway and is something that can be produced from time to time, i agree.

For each graph there is a different gfex file and it changes the coloring of the nodes beside the shape of the graph, you can find the files in here:

https://github.com/AlbertoIT/Merit-Network-Visualization/tree/master/data

[DdmrDdmr]

<...>

Ok, I saw that the size of the node was capped to 500 in the OP. The size of the node is therefore overall, as a sMerit sender, not relative to the sMerit sent to me specifically, but in general Forum wide If I understood it properly now.

[Piggy]

Yes is working that way, i just added the rule i used to show how the parameter which govern the size of the node is calculated:

Code:

 If Merit Given > 500 
		Node Merit Size = 500
	   else
		Node Merit Size = Merit Given

For example From the User Information Panel:

TMAN
Merit Received: 601.0

Merit Given: 1590.0

Merit Given + Received: 2191.0

Rank: Hero Member

Trust: +1

Node Merit Size: 500.0

babo
Merit Received: 43.0

Merit Given: 139.0

Merit Given + Received: 182.0

Rank: Legendary

Trust: +1

Node Merit Size: 139.0

Another detail i would like to point out.
The goal of this work is to find new merit sources, but if you look at the peripheral area of the graph you can see there are a lot of isolated graph, i'm not 100% sure, but there could be a lot of abusers there. If anybody want to investigate, would be relatively easy since there is a version of the graph where red trusted users are already marked.

[Piggy]

Really neat, even though user with slow connection like me need to wait for a while (10-60s)

It is quite a lot of information, in fact the waiting time in the begin goes mostly in the downloading of the file.

There are exactly 15954 Nodes(Users) and 54108 Edges("Merit connections") in there. If there are more merit transactions from A to B they will be grouped together and be represented with 1 line that goes from A to B, basically a "Merit connection"

[athanz88]

Wow this is a nice visualization of user's merit. What i like the most about your work is, it can be used to look on how worthy a person is, based on the big names who gives them merits. I know that the weight of the merit is the same across all the member, but earning merits from big names sure give you some kind of "proud" feeling over it.

And it is easier to notice that via this visualization, rather than scrolling through a member's merit history list.

[vlad230]

This looks very interesting

It would be nice if we could use a system like this to see a user's post distribution across the boards for a better view of one's activity here on BCT.

For example, we could see which users mostly post in bounty threads

[Piggy]

Wow this is a nice visualization of user's merit. What i like the most about your work is, it can be used to look on how worthy a person is, based on the big names who gives them merits. I know that the weight of the merit is the same across all the member, but earning merits from big names sure give you some kind of "proud" feeling over it.

And it is easier to notice that via this visualization, rather than scrolling through a member's merit history list.

The idea of this kind of this visualizations is indeed to give an immediate visual impact (and expose patterns ) through simple properties( size of the graphs, colors, size of nodes, connections ) in order to have an immediate idea of what is happening, without having to look and analyze raw numbers.

This looks very interesting

It would be nice if we could use a system like this to see a user's post distribution across the boards for a better view of one's activity here on BCT.

For example, we could see which users mostly post in bounty threads

This is more useful to represent interactions between user, but something can be already done on a similar topic, check in which board a user is awarding merit mostly

https://albertoit.github.io/Merit-Explorer-SQL/

Code:

SELECT SUM(Merit),Board,SubBoard 
FROM MeritData 
INNER JOIN UserData on fromID=UserData.UserID
WHERE UserData.Username like "vlad230" 
Group BY Board,SubBoard 
ORDER BY Board,SubBoard;

in here you can find other SQL query examples if you are interested: https://bitcointalk.org/index.php?topic=4551881.0

Really neat, even though user with slow connection like me need to wait for a while (10-60s)

It is quite a lot of information, in fact the waiting time in the begin goes mostly in the downloading of the file.

There are exactly 15954 Nodes(Users) and 54108 Edges("Merit connections") in there. If there are more merit transactions from A to B they will be grouped together and be represented with 1 line that goes from A to B, basically a "Merit connection"

I see, but i think it would be better if there's simple loading bar/percentage. At first, i thought your site isn't compatible with my browser or my security extension break the page.

You are right, i have been working so long with it i stopped seeing this sort of problems, i'll make it clear somehow that is loading the data in the begin

[Piggy]

i now made a change which should help understand the data is actually being loaded and everything is "fine", for those who may wonder why they see an empty screen for a while before anything happen 

[lequocvuongpro]

This has been quite a long and challenging project given the amount of information to show, i have started to look into it since i saw the network graph threads of DdmrDdmr and sncc. My goal was to have an explorable version of the full network with interactions included, as well as my take on trying to find new merit sources.

Each node represent a user and the edges with arrows merit transactions, received or given.
I made 2 versions, the main difference is in the shape. Then for each of these versions i made two variants,one where the coloring is based on the Rank and the other is based on the Trust.








The node size is based on the Node Merit Size variable (on the User Information Panel) and reaches the maximum with a total of 500 merits given away, i did this to try to avoid the current merit sources shading emerging one. This should help visualize at sight the main givers just by looking at the size of the nodes.

Node Merit Size is calculated in the following way:

Code:

 If Merit Given > 500 
		Node Merit Size = 500
	   otherwise
		Node Merit Size = Merit Given

The node and edge color simply tells the rank/trust of the node.

I removed from the network all the nodes and edges where:

Code:

Merit Received + Merit Given = 1

So anybody who sent a single merit or received a single one won’t be in it.

There are exactly 15954 Nodes and 54108 Edges in there. If there are more merit transactions from user A to user B they will be grouped together and be represented with 1 line(Edge) that goes from A to B

What is the purpose?

You can use these visualizations to help identify users which could be a merit source, since at first glance you have a very intuitive way of knowing the amount of Merit given (size of the node) and Rank/Trust (color of the node). Plus the personal network tells the merit story of the user.

How i made this?

I prepared all the data i needed and made several iteration with Gephi for the layout of the network graph. The drawing is made using the library Sigma Js which supports WebGL.

How to use it?

• Each node you see can show the user name on mouse over and by clicking on it it will isolate the connections based on merit sent and received, both in the form of a graph and with a panel, which shows supplementary informations about the use selected as well. Moving the mouse over the names on the list will show the label on the graph as well.
  
• You can drag the network by clicking on an empty space, zoom in and out with your mouse wheel or using the buttons in the bottom overlay. The magnifying lens with the square center the graph on screen.
  
• On the left panel you can search for any user by name.

Links to the visualizations:

https://albertoit.github.io/Merit-Network-Visualization/
https://albertoit.github.io/Merit-Network-Visualization/?config=config_Trust.json
https://albertoit.github.io/Merit-Network-Visualization/?config=configsparse.json
https://albertoit.github.io/Merit-Network-Visualization/?config=configsparse_Trust.json

This very great idea for Bitcointalk, Hope it's will be applied for Rule of Merit rank. Thank for your sharing. Good luck !!.

[nngella]

May I know the reckoning date of the data that you used? 
I just search some users and I think the data is not updated.

You can see merit abusers in this by looking for trading of merits or giving abundant merits (also to identify multiple accounts)

[Piggy]

May I know the reckoning date of the data that you used? 
I just search some users and I think the data is not updated.

You can see merit abusers in this by looking for trading of merits or giving abundant merits (also to identify multiple accounts)

This is using merit data available till 1st of July, i may update it later on.