1523 jpg
(1.78 MB, 3287x4111)
>>/8413/
>>/8414/
It maybe barebones but I am still pretty impressed with it.
> Overall, I'm not quite sure if this can be as useful as I imagined. Maybe if it was executed in some fixed intervals with proper actions on match (like download of the matched item) it could be somehow useful, however I'm afraid of way too big number of false-positives.
I think even as a simple aid in tracking it is useful already. At the very least, I could imagine running it with filters for mentions of places and being able to get at a glance if something is going on.
How powerful it could be depends on several factors I reckon. Twitter and 4chan with mid level filtering might be perfect for being able to react as soon as something occurs considering past controversies that come to mind and how they played out. If stuff goes down on a site that lacks RSS or is hard to track that might be a different story; say locked behind the chaos of Discord server without the ability to use this tool at all (say, Knighty being pressured on Fimfic's discord by a group of users and caving quickly, not saying that would happen, but as a example of a lot of sites having Discord circlejerks were moderators are at and a lot of drama can play out). Also:
> Also there will have to be more research on the keywords we should look for, best thing would probably be to get lot of links to posts about bad happenings so they could be analyzed.
Agreed. If we could narrow down certain keywords that would be useful. My instinct tells me something like this would be a start:
places = ["4chan", "Derpibooru", "Derpi", "Twibooru", "Everypony.com", etc]
triggerwords = ["purging", "banning", I can't think of much here atm]
< or if twitter
= [Nazi, Racist, Tankie (yeah, people forget but at one point Derpi even had the position of removing communist Umbridge memes and I remember a handful supporting it), Racism, I can think of a fair bit but you get the picture]
Though something like this directed at twitter would get a lot of false positives, if it was set at the right people (obviously, people involved with MLP) a fair bit I imagine would be relevant even in this more basic form. Still, testing has to be done before I can say that with confidence.
> BUGS - [NonBreaking] Java 11 doesn't like one Clojure built-in function, so there is one error message about reflection, however program should work fine. This happens only with java 11. Other tested versions (8, 16) seem to not mind.
I did encounter this bug as openjdk11 is what I have installed atm when I tested this. Did only a moderate amount of testing though, hopefully soon I will get more time to really test this out.
> As it's based on an older code base, language used is Clojure, which is Functional Lisp dialect for JVM. At first I wanted to rewrite it in something else but then I tried it and it is actually quite nice when you get used to it, so I'll stick with it for now.
Perfectly fine. I will be even less of a help than I was with Python on the codding side of things. Not sure if I'll get anywhere but maybe if I get some free time I might try to teach myself this and get myself at a level of being barely helpful!
> Anyway, I'll keep working on this along with the comment converter and see how it goes.
As I always say, do so at your own pace and don't stress on any deadlines.