LinkedIn, Please Take on Group Spammers

Almost like clockwork, I get a set of daily LinkedIn notices of off-topic job listings improperly posted as discussions to a Linkedin group I belong to. They’re spam, and they drown out genuine discussions. They degrade the value of the LinkedIn group where they appear. The group owner is either disengaged or doesn’t care. He ignores requests to remove and ban the frequent-repeat offenders, a coterie of accounts that match a couple of patterns that are easy-to-see, that is, if you’re looking. I’m hoping to get LinkedIn will act where the wayward group owner won’t, hence this blog post: A call to LinkedIn to please take on group-discussion spammers. Systematic analysis linking group-posting behavior to account configurations will do the job:

Data science applied to quell a certain variety of abusive spam.

LinkedIn, here’s what I see. Start with a bunch of Text Mining Group discussion spammers. These people have lots of disjointed group memberships but only minimal work-history listings, a few jobs with no descriptions:

The job titles, employers, and group memberships of each of these LinkedIn profiles are not consistent. A real-person’s profile will have some linkage between what a person does, who that person works for, and what groups that person belongs to. And some of the employer companies listed are almost surely bogus. Thin Coal SpA, given in the last of the profiles above? Give me a break. So look for accounts that lack internal consistency and you’ll find likely spammer accounts. Confirm by looking for spamming behavior — bogus companies are a further confirmation — then kill the accounts.

Other LinkedIn accounts indulge in spam group discussion postings but have more extensive profiles than the seven I listed above. They also share the trait that they belong to a grab-bag of disparate groups; perhaps they spam groups other than the one I belong to, Text Mining. I mean, check it out:

Interesting job descriptions that “Maria Sanchez” has. Copy-and-paste one sentence from one of those job descriptions into the Google search box, “Provides input and recommendations to the Talent Acquisition team for strategies and initiatives.”, and up pop four LinkedIn profiles that use the same text:

If I search within LinkedIn on the same text, I see eighteen profiles that text, and remember, I see profiles only of accounts within my network. No doubt there are many more potential hits that I can’t see.

Clearly someone created those accounts via a copy-and-paste job. Had I LinkedIn’s resources, I’d systematically compare descriptions across profiles to find similarly cloned profiles. At best they’d indicate real-person plagiarism, perhaps forgivable, and at worst, they’d suggest an account set-up by a spammer. Again, suspicions should be confirmed by looking at actual use of the accounts.

LinkedIn, how about it? How about booting those spam accounts I’m pointing out here?

And readers, if you’re into text mining, rather than the Text Mining group, I recommend the Text Analytics group. I’m a manager, but regardless, the group is bigger and it’s largely spam-free. And also check out the Sentiment Analysis group. I’m partisan, but if you’re into language-technology networking, these are great online venues for you.

Leave a Reply