If you’ve used Twitter for any length of time, you’ve come in contact with its massive underbelly of spam. Today, the company shared details of its BotMaker anti-spam system, which has contributed to a 40 percent drop in Twitter’s own spam metrics since it was introduced.
As Twitter notes, one of the difficulties it faces in combating unwanted content is that spammer “know (almost) everything” about its counter-measures because of the developer API. The real-time nature of Twitter also means that any systems must operate without adding latency to the service.
BotMaker is governed by three guiding principles, according to Twitter: “prevent spam content from being created, reduce the amount of time spam is visible on Twitter [and] reduce the reaction time to new spam attacks.”
The system consists of a collection of rules, which are referred to as bots, that govern what to do when an action is flagged. Its processes are split into three different timings: real-time, near real-time and periodic (offline).
The rule syntax for BotMaker is designed to be human-readable, modular, and quick to deploy. The goal is for Twitter to react to spam attacks within minutes, instead of hours or days as before.
Here’s a graph showing the drop in spam activity on Twitter after BotMaker took effect:
There’s certainly more to be done on Twitter’s part in order to eliminate spam, but its latest efforts appear to be having a positive effect. Click the source link below for more technical details about how Twitter accomplished the feat.
➤ Fighting spam with BotMaker [Twitter Engineering blog]