So I was a little bored this evening and for a reason I already can't remember, I decided to make a python script to create visual representations of the implication chain/tree for any tag.

Here's the result for vore (149kB)
Here's the result for cum (1.14MB)

It's still a bit buggy (the example for cum has a lot of double and triple arrows) and the layout is certainly not very easy to read, but I think there might be some use for something like this in organizing tags on the site and keeping an eye on long forgotten aliases (like how cum_in_cup still has a pending alias for cum, despite having been implicated to cum_in_a_cup (which does not imply cum) 4 years ago).

Is there any interest in me continuing something like this?

Honestly, I think this is pretty dang neat.

I'm not precisely sure HOW it it could be used, but as a visual person, this is very appealing.

Goodness knows I spend a lot of time tracing implications and aliases around though. I would +1 your continuation :D

SnowWolf said:
Honestly, I think this is pretty dang neat.

I'm not precisely sure HOW it it could be used, but as a visual person, this is very appealing.

Goodness knows I spend a lot of time tracing implications and aliases around though. I would +1 your continuation :D

Heh, thanks. I figured you in particular would find some use in graphs like these as visual aid (I got the idea from reading you say how the implications for vore were kind of a mess), so it's nice hearing you say that.

Now I mostly just need to figure out how to make it conveinient to use without having to set it all up (not that the setup process is all that complicated, but most people probably wouldn't want to deal with it anyway) and, most importantly, to make it fast, because the graph for cum took something like 20 minutes to generate because of how little data I get from the API routes for implications and because I'm respecting the 1 call/second rule.

Fifteen said:
Heh, thanks. I figured you in particular would find some use in graphs like these as visual aid (I got the idea from reading you say how the implications for vore were kind of a mess), so it's nice hearing you say that.

Now I mostly just need to figure out how to make it conveinient to use without having to set it all up (not that the setup process is all that complicated, but most people probably wouldn't want to deal with it anyway) and, most importantly, to make it fast, because the graph for cum took something like 20 minutes to generate because of how little data I get from the API routes for implications and because I'm respecting the 1 call/second rule.

you already know me well ;) (I'll reply to your messages later, been busy though!)

I would also be concerned about readability, but I'm not precisely sure how to improve that. My first thoughts wander off to... well, way more complicated collapsable boxes and whatnot haha. but that's basically way over complicating this.

I have really nothing helpful to offer, as I know nothing of these things :D

Try setting the graph attribute ratio. You can make the generated output diagram be wide (r < 1), tall (r > 1) or square (r=1). Square or slightly wide seems like it would be appropriate here.

You might also be able to set the digraph as strict -- this is certainly possible in .dot files, and would prevent your duplicate edges problem (not sure whether it would be by raising an exception or simply ignoring duplicates)

TBH the many threads for implication/alias discussion is a robust but disorganized design.
So I think this could be a pretty useful tool with a little more work..

One question is, whether you could render to HTML like dot + co. can? If so, then the idea of including links to threads and perhaps wiki seems like a fairly obvious improvement (headhref/tailhref on edges, href on nodes)

savageorange said:
Try setting the graph attribute ratio. You can make the generated output diagram be wide (r < 1), tall (r > 1) or square (r=1). Square or slightly wide seems like it would be appropriate here.

You might also be able to set the digraph as strict -- this is certainly possible in .dot files, and would prevent your duplicate edges problem (not sure whether it would be by raising an exception or simply ignoring duplicates)

TBH the many threads for implication/alias discussion is a robust but disorganized design.
So I think this could be a pretty useful tool with a little more work..

One question is, whether you could render to HTML like dot + co. can? If so, then the idea of including links to threads and perhaps wiki seems like a fairly obvious improvement (headhref/tailhref on edges, href on nodes)

All good ideas. This is the first time I've ever used graphviz, so I'm still stumbling around a bit.

SnowWolf said:
I would also be concerned about readability, but I'm not precisely sure how to improve that. My first thoughts wander off to... well, way more complicated collapsable boxes and whatnot haha. but that's basically way over complicating this.

I have really nothing helpful to offer, as I know nothing of these things :D

I initially wanted to do more of a circular graph, but I tried that and the result wasn't a while lot more readable, not to mention I can't group aliases under big boxes like I can with the attempts I linked earlier (it doesn't seem to be a feature, anyway). I'll stick to boxes for now.

Regarding the collapsible tree thing like they have in some places to browse directories and whatnot on some websites and file explorers, that could certainly be done with the data we have, though I'm not looking for something too interactive yet.