![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
MATT1993:
(clips of everything stupid he has ever posted on LJ/DW play at once, indistinct)
So lately I've been thinking about figuring out for sure which entries are the most cringeworthy and then making those only accessible to long-time readers, or something like that... ...make a program that takes that and analyzes it to give each of my entries a "cringe score" of sorts... Then I'd use that to divide my past entries into "cringe levels" from best to worst, and then I'd adjust the visiblility of my entries based on that.
But it's over two years since then and I never got around to finishing that because, for one, it would take forever to do at the rate I'd been going. BUT! Apparently there's now AIs for detecting toxic comments online!
I think it's really, really, REALLY great that people are starting to research machine learning for detecting toxic comments online so that they can be filtered out automatically and/or so that people (like me!) can learn what makes their comments potentially toxic and how to avoid it. You know what would be even better, though? If I could actually use it. Finding a website that lets me do that is harder than I thought...
It's been another year and a half since then and most of the toxic comment detection AIs that did exist have now become defunct or become members-only, and it doesn't look like there's been a lot of new ones. The one I was just barely able to use before - PerspectiveAPI - seems to no longer have a public demo available and even if it did, it was geared more towards rating comments. Meaning I couldn't use it on entries in my journal longer than the average comment - the best I could do is put in individual sections of the entry, rate those, and assume the rating for the whole entry would be close to the average of the individual ratings.
But I have so much anxiety about certain things I've said in the past - and it's gotten much worse this year - that clearly I have to do something about all my past entries and comments. I might just make all of my entries before I started posting about Character.AI friends-only for now, even the ones I feel are okay, and then reevaluate once there's finally another AI that can rate each individual entry and comment and that I can actually use.
I ask you: Is there another tool available that can detect toxic comments online where I can either paste them in one by one or (better yet) have it go through every entry and comment in a given LJ/DW journal automatically? In fact, I'm even making this entry public (at least for now) just to increase the chance that someone who has an answer will see this. (And now I'm realizing the last time I asked about this, I also made that entry public for the same reason. Go figure.)