Perspective API - help needed
Did anybody successfully integrate the Perspective API and can lend me a hand on what I'm missing? I generated successfully an API key and filled it into the respective field in Studio settings. In google console there seems to be an activity when testing toxic, but it doesn't seem to integrate with the frontend. What's the trick?
-
@Alex T⚜️ , can you help, please? I've got the feeling that registrating for an API key and inserting it into Studio settings is not the end of the story.
-
Did you enabled the toxicity filter itself? Did you specified the type of action you want it to take with a toxic content? I mean the settings at the [studio -> antispam -> toxicity filter].
Also what exactly do you mean by "it doesn't seem to integrate with the frontend"? It should take action upon any form submission. So for example it should prevent you from posting a toxic post/content in case the "Action" setting is set to "Block all toxic content".
-
Hi Alex, thanks for taking care of my problem.
Yes, the toxicity filter is enabled. I selected "Disapprove all toxic content if possible, block the rest" as the action to take.
The problem is that trying it on my test server nothing gets ever blocked, even when I'm sure the text is more than toxic. The admin doesn't get a report, though set in settings.
The only thing I can confirm is that the diagrams on the google cloud console show activity when I save toxic content in forms.
-
Have you checked the status of a content you're posting? If you set it to "Disapprove all toxic content if possible, block the rest" then it allows you to post it but should make it suspended to be reviewed by admins later.
Also don't use an admin account for testing, because it is having more privileges.
-
Ok, thank you for the hint. If I now try to send a comment with toxic content (threshold 70%) as a user without admin rights, the Post button gets a spinning wheel and the content cannot be published. But how does the user know that it's due to the toxicity in his text?
-
What mode you're testing it in? Also what is your UNA version? On a latest UNA release in the "Block" mode it works as expected by preventing a form from being submitted:
-
Our test server always runs on the latest available version of the UNA framework for testing purpose, currently it's on 14.0.0-RC3. Mode is, as mentioned before, "Disapprove all toxic content if possible, block the rest", Toxicity thresholde is 70%. Now that I changed the mode to "Block all toxic content" in order to check if I get the same result as you, the system now doesn't block anything anymore.
-
Then please ensure that you're not testing it using an administrator account and that a content is above the toxicity threshold. Checked it a minute ago on a latest version and it works fine for me (see screenshot above).
-
Ok, I'll see if it will start working, again. Maybe a total block is the safer option. When I tried it in mode disapproving toxic content, if possible and blocking the rest, I didn't get any report from the system and just a spinning wheel in the post button, as demonstrated in my screenshot.
-
It works for me now in mode "Block all toxic content", too. Toxic content can't be published, but no reports to the admin in this mode. I guess, that's only for mode "Disapprove all toxic content if possible...", which doesn't work for me. Anyway, a great help to protect users from bad content.
-
Email reports are working now, too.
-
@Alex T⚜️ , the problem came from different mode settings in my root instance and in my test instance, running in a subfolder of the root. Since I used the same API key and had both active, it confused the perspective system and it couldn't send a response. As Andrey said, 99 percent of support requests stem from user errors...
I get the reports now, but only on my system admin email account. Is there a chance that you could implement reports for moderator members, as well? Maybe as an addon to user reports in Studio Dash?