I posted as new thread before I saw this thread…
Tool for Q-searchers. Some fragments mapped, ball is in your court.
https://mega.nz/#!WfhRlASA!gaDqWNdpP-OSWY4bue0Oj0En8h9lXX85J0FFIXytRGI
Node JS, Electron, Cytoscape, GunDB
cd into extracted directory and run "npm run"
GunDB lets you have property lists for each node and many connections. Q Analysis app lets one manage those individual properties. For each node with only name and image displayed. Each node can cross link to another graph file or be linked to a url. All html css js. Can customize presentation layer.
Yes on all points. However, for my part, I’ve got too little time to maintain my self. Source is there. Any one can read and decide if they wish to utilize it. Its a start, hopefully others can help realize it. I can achieve what I set out to be able to achieve and figured others might get something out of it.
AI can assist in this. I’ve been working on AI to monitor reddit for shills and abusive moderation. Talking point detection is on my list too.
If we can detect these things with AI we can respond to shills with evidence to refute their argument and with moderators, archive and record events to expose the infiltration.
I’m hesitant to post code yet, not sure it’s a good idea to put something out that can be weaponized by trolls the same as we use it as defense.
Right now I do an ok job of detecting when people are arguing vs debating, asking questions vs concern shilling. Starting to narrow in on detecting posts likely to be removed by a boards moderator but need more sample data.
None the less it’s possible. We just need great training data. Sample conversations with shills and sample moderated posts.
I’ll post source this weekend. Real life has been occupying free time.