Scripted moderator tools?

[TheLordOfChaos201]

ok how about we organise a small team of programmers to design a tool for moderating models.

that way even a noob can moderate a model and it wouldn't be their say but rather the moderating tool that points out the flaws.

this would decrease the number of poor entries and perhaps even influence the model makers to use those moderators tool to help evaluate their own maps.

.... I'm busy for the next 2 months, I won't be able to create my Map moderating tool until that's over.

 

[JesusHipster]

that way even a noob can moderate a model and it wouldn't be their say but rather the moderating tool that points out the flaws.

I don't think this is possible, really.

 

[TheLordOfChaos201]

I don't think this is possible, really.

it cann't be impossible so there for it must be possible

 

[JesusHipster]

it cann't be impossible so there for it must be possible

That's a good logic, but unfortunately it is indeed impossible. First off there is an artistic aspect to every model that a script can not be taught to consider. Secondly what kinds of flaws can tool point out? Most of them are visual - ex. attachement point having a bad position in space, wrap being shit and so on. This isnt something a tool could point out And also, consider that there are different kinds of models, like UI, effects and so on.

A script could possible point out if the model makes wc3 crush, all though it is a rare problem already.

 

[TheLordOfChaos201]

well a programmers could probably check if the bones of the model all move
the skin isn't broken
there is no broken models
check to make sure it's not a warcraft standard model..

 

[JesusHipster]

well a programmers could probably check if the bones of the model all move
the skin isn't broken
there is no broken models
check to make sure it's not a warcraft standard model..

Not all bones are supposed to move on all occasions

Broken ones are pretty easy to spot anyway and get uploaded pretty rarely.

Writing a tool like that is an effort spent in vain.

 

[The_Silent]

Well, a modeling moderation tools that simply points out missing attachment points, animations, ect. would be quite simple, but it would only be an aid to the moderator, and could never replace them. It would be most effective to have it build into the website, so the site simply makes a list of possible bugs whenever a moderator starts reviewing the model.

Like:"

There is possible issues with this model.

• No death animation.
• Lacks origin attachment point.
• Birth animation is set to looping.
• Positive extends contains negative values.
• Ect. ect.

"

But that is it really. Metrics alone can only get a model rejected, not approved. (Since a model might still be bad and have no obvious issues, but not be approve-able and lag the basics).

Also, doubt other resources has so easy to measure metrics.

 

[MasterBlaster]

But maybe that would be enough? I mean, a program that automatically rejects the models that have some of the most common issues?

 

[The_Silent]

I'm per definition against "automatic" rejection. But as a helping tool for the moderator it could be useful.

 

[Ralle]

We'd need some pretty neat neural networks to decide what is useful and what is not bugged. Good luck. See you in 5-10 years when you have a semi-working prototype.

 

[A Void]

time to make Artificial Intelligence...

 

[Rui]

time to make Artificial Intelligence...

Exactly. @JesusHipster I wouldn't say it's impossible, but the expertise and time required to program such algorithms is well beyond any of us, not even considering how their complexity would weight on the server and how long it'd take; users could be stuck for quite some time after submitting models (considering most have at least the base model and its portrait).

Simpler programs to check basic stuff are possible, but I don't know how much they could. It shouldn't be a priority IMO.

 

[TheLordOfChaos201]

it can't be that difficult... granted I know almost nothing about models, it should still be possible to do something

I've recently learnt how to data mine... if there is a common problem between multiple models, I would be able to identify it

of course all I've been able to do with it so far is extract 3500000 definitions for words over the net

 

[PurgeandFire]

I'm per definition against "automatic" rejection. But as a helping tool for the moderator it could be useful.

This.

If it was added, I would just have it as a helper tool for mods. Little "recommendations" or reminders of things. But they'll still be judged on a case-by-case basis.

e.g. poly counts and particle counts can vary wildly by model. If it is a tiny effect, then the limits will be more stringent. If it is a massive complex model, we'll usually be more lenient about the counts.

It could be pretty useful for the spells section. A lot of the changes we recommend are repeated from other reviews (functions to avoid, missing tooltips, etc.), or the maps section. So it is a cool idea, but I'd only use it as a helper tool.

 

[TheLordOfChaos201]

____
Ralle!!!!
____

Do you know if there is anywhere on the Hive with a lot of warcraft item names, history, spell names, hero names.... and so on

I've recently learnt how to data mine, and I would like to data mine the hive for these uniquely warcraft words in order to create a map moderating tool

Right now I'm in the spell check phase, but I hope to also create a system of checking code for common errors