- uıɐbɐ ʎɐqǝ ɯoɹɟ pɹɐoqʎǝʞ ɐ buıʎnq ɹǝʌǝu ɯ,ı
- From: New Zealand
- Registered: 2008-04-16
- Posts: 6,680
OK, Apiv2 - let's discuss this
OK, pile in (and be respectful), what features / options / changes do you want to see in the APIv2. Please give more than just a "add widget", and please give some background.
- Registered: 2008-10-09
- Posts: 1,313
Re: OK, Apiv2 - let's discuss this
You want me to remember all the suggestions I've made over the years? I'm sure they're all posted on the forum at one time or another... I'll have to think of what they were again...
Also, when you say API, do you mean the query API, or submission of spammers as well (as that's also an API, albeit a different one).
One that comes to mind though, is some sort of AI based on API queries... say for example that a spammer/bot starts hitting a bunch of forums with a fresh set of credentials (email, IP and username), and forums being hit start querying the API with those credentials, the SFS server could detect patterns (such as multiple forums querying the same credentials over a short time period) and detect this as potential bot activity that could be blocked proactively, even before the data is actually submitted... this could be done by having the confidence score go up based on number of recent unique queries for the data, even if it's not submitted yet. The DB could even be fed automatically via this mechanism... if more than X unique IPs query the same data over a time period of Y, automatically add the data to the database.
This would allow new spammer credentials to be blocked even before the first submission is made.
On a similar vein, have the lastseen/frequency/confidence on submitted spammers reflect recent queries as well... that way if submitted data sits unused for a while, but then the spammer starts using it again, the lastseen will reflect this after several forums query the same data again... helps keep data current even if the spammer hasn't used it to spam yet/hasn't been submitted yet. To prevent abuse/submission of false data (submitting trolls for example), require a minimum number of hits by separate forums/IPs before it is reflected in the data.
Last edited by kpatz (2018-05-21 7:37 pm)
We need better spam proofing technology - preferably the kind that electrocutes spammers with 50,000 volts 100 amps the moment they click "Post".