Community Network Visualisation

merefield · April 5, 2021, 5:16pm

Summary: A user network visualisation

GitHub: https://github.com/merefield/discourse-user-network-vis

Install: Follow the plugin installation guide.

Features

Presents your community as a network visualisation (Desktop/Tablet only).

Get to it by hitting User Network on the Hamburger menu or navigating to yoursite.com/usernetworkvis - NB you have to be logged in as a registered user to see the vis.

the nodes represent users.
the links represent interactions on your forum - the more interactions, the thicker the line
- an interaction is the combination of Likes and Replies (this is the very same data that’s available on someone’s Summary profile page to all users)
Hover over a user to isolate them and their connections visually
Click on a node to navigate to their Summary profile.
Drag nodes around and watch the show

Trust levels are currently exposed as different colours. If you don’t want this, set the same colour several times. I may find another criterium to set colour, but this was the most obvious by far as very generic

To refresh data, Trigger the job. This may be necessary after a settings change (I’ve marked those that require it):

Go to: https://your.site/sidekiq/scheduler and find this one:

Otherwise this runs every hour.

if you have 10,000’s of users, this job may run LONG and take up resources. Seems to work ok on a 5,000 account site running on a 2 Core 2GB VPS taking about 30 seconds. The SQL is necessarily BRUTAL

Interested in the data?

My intention was with this was just to paint a picture.

If you want to see the data, just go to yoursite.com//usernetworkvis.json

In any case, in individual cases, you can see much of this data on each users summary page by clicking on a node (Most replied to, Most liked/liked by).

Configuration

Settings … so … many … settings:

If you have a large site with lots of people on it, you are going to find the diagram very noisy. Consider upping minimum trust level.

CHANGELOG

we’re at V 0.3, it’s pretty early days.

TODO

Suggest away

Creds

The visualisation is based on the awesome d3 library

Similar projects

If you like this you might also like Word cloud plugin for discourse? - #4 by merefield, Discourse for a "digital garden" - #5 by merefield

awesomerobot · March 24, 2021, 1:36am

Very cool! This isn’t exactly the same thing, but it reminds me of a study I read about a few years back:

merefield · March 24, 2021, 10:31am

Oh nice find, I wonder if it will drive any engagement? … Same presumably might be said of Who's Online Plugin (discourse-whos-online)?

You know that gives me another idea: you could put “halos” around nodes that were online perhaps leveraging that plugin if available …

cogdog · March 30, 2021, 7:18pm

Thanks for this, just installed for an educational community where I’m still trying to budge up the participating level. We get lots of signups for event related activities, but it goes too quiet in between.

I set Minimum Trust level to “2” to show the more active users w/o a lot of noise.

cogdog · April 5, 2021, 4:28pm

It could be helpful to have an option or view of the tabular representation of the data.

merefield · April 5, 2021, 4:51pm

My intention with this was just to paint a picture.

if you want to see the data, just go to yoursite.com//usernetworkvis.json

In any case, in individual cases, you can see much of this data on each users summary page by clicking on a node (Most replied to, Most liked/liked by).

cogdog · April 5, 2021, 4:50pm

And it’s painting beautifully. Users of plugin always want more, we can dream. Thanks for the link to the json.

merefield · April 5, 2021, 4:55pm

You might also be able to adapt the SQL in the query relatively easily and run it in the Data Explorer:

github.com

merefield/discourse-user-network-vis/blob/55354d641deba7a4a93670a36488abee9a7115c2/jobs/user_network_stats.rb#L15



def execute(args={})

  user_network_list = []
  user_network_link_list = []

  min_tl = SiteSetting.user_network_vis_minimum_trust_level

    build = DB.build <<-SQL

    select source_user_id, target_user_id, sum(score) as score from (
      (SELECT user_actions.user_id as source_user_id, user_actions.acting_user_id as target_user_id, count(*) as score FROM user_actions
      INNER JOIN topics ON topics.deleted_at IS NULL AND topics.id = user_actions.target_topic_id
      INNER JOIN posts ON posts.deleted_at IS NULL AND posts.id = user_actions.target_post_id
      WHERE topics.deleted_at IS NULL
      AND (topics.archetype <> 'private_message')
      AND topics.visible = TRUE
      AND (topics.category_id IS NULL OR topics.category_id IN (SELECT id FROM categories WHERE NOT read_restricted))
      AND user_actions.action_type = 2 GROUP BY user_actions.user_id, user_actions.acting_user_id ORDER BY COUNT(*) DESC)
      UNION
      (SELECT posts.user_id as source_user_id, replies.user_id as target_user_id, count(*) * #{SiteSetting.user_network_vis_reply_multiplier} as score