False positive spam flag resolution

Curators
1

On iNaturalist, “spam” is defined as commercial spam…

…anything that is clearly intended to make money, which could be linking to spurious sites, or trying to manipulate search engine indexing through lots of links to weird places…

…so it excludes things like observations of humans or inanimate objects. That said, some users find those types of observations stupid/pointless/annoying, so they often get flagged as spam. Additionally, the Akismet spam-detection system doesn’t always identify spam correctly, and sometimes marks otherwise innocuous non-spam content as spam. See list of resolved spam flags. Flagging as spam hides content from other users, and if someone gets too much of their content flagged as spam, they’re automatically suspended. :frowning_face:

The most common types of content that are flagged as spam, but usually aren’t spam are:

Check the flagged content, and if it’s not actually spam (iNat’s definition), then resolve the flag, marking it as “not spam”. If it is spam, for now the process is to just leave it unresolved.

Also check the user’s profile to make sure they weren’t marked as a spammer. To unmark a user as a spammer, click Admin tools in the lower left, then Flag as non-spammer. After you flag them as non-spammer, send them a message informing them that their account has been unsuspended. This is the message I use:

Account UNsuspended

Hi! Unfortunately you were the victim of iNaturalist’s overactive “spam-flagging bot,” so your account was flagged as spam and suspended. I’ve unsuspended it and marked you as not a spammer.

8 Likes
2

So I see some comments that are clearly inappropriate, but don’t necessarily meet the “money making” criterion for spam. Should we be putting effort into reclassifying these, or not bother?

1 Like
3

I would use spam flags to flag anything that is not appropriate, until another measure is given. If something is obviously not appropriate, I think curators can also delete comments themselves.

1 Like
4

I’ve seen a lot of these flags as the CNC goes on, so something to be aware of.

2 Likes
5

We’re working on a hiding comments (and other posted material) functionality, which I think @bouteloua (?) requested a while ago. Once that’s implemented it should be a good way to hide offensive non-spam material but still keep a record of it so we can evaluate if the user requires suspension.

5 Likes
6

Can a couple other curators or staff put checking these flags/suspended users on their repeating to-do lists?

I just unsuspended a plant taxonomist who had been super helpful with adding IDs. He had been suspended by the iNat spambot for a few weeks. :(

1 Like
7

I’ve found a couple of these recently too but also one that I thought was something like this and turned out to be a valid suspension due to bad behavior. Which is another reason the documentation is valuable.

I think we are moving in the right direction in dealing with iNat website moderation vs taxonomy curation, but it is definitely important and a slow process, and can get tricky.

1 Like
14

@kueda Would something like having 10 improving IDs be a feasible option to help prevent IDers, those who have few observations, from being flagged as spam and their accounts suspended?

4 Likes
15

Sure. Any reason not to make it 3 improving IDs? Currently the system assumes you’re not a spammer after you’ve made 3 RG observations. Do you think the ID threshold should be higher b/c it’s a bit easier to get improving IDs?

1 Like
16

If the status of those improving IDs changes with further community IDs, would the system assumption also change again?

17

nice! I don’t have any strong opinions on the number of IDs.

1 Like
18

No, once the system thinks you’re not a spammer it stops automatically checking your content to see if it’s spam (but you can still be marked as a spammer if actual people explicitly mark your content as spam).

2 Likes
19

I don’t have a strong opinion on the number of IDs either, but I think this is a good idea to try. If it ends up being bad (ie allowing spammers in) then it probably wouldn’t be too hard to roll back?

20

I wonder if since there is already a threshold for sending direct messages (three verifiable observations or IDs for others), that the system shouldn’t auto flag any messages as spam anymore? There are many that need review by staff, though I can’t really tell how good or bad it is at detecting actual spam.

1 Like