Unsolicited advances are a problem on all social platforms, but they seem especially out of place on LinkedIn, due to the platform’s professional and career focus.

And yet, many people – many women in particular – do indeed experience harassment on LinkedIn.

Last year, former LinkedIn CEO Jeff Weiner addressed the problem publicly, and said that they were looking to take more action to stamp InMail harassment out, yet Weiner also declined to provide numbers on how often such harassment was occurring. But there are various accounts online of serious issues with LinkedIn harassment. And now, LinkedIn has a new system to detect and reduce the impact of such for its users.

In a new post on the LinkedIn Engineering Blog, LinkedIn outlines its new detection system which has been trained on examples of past reports of harassment via LinkedIn messages. 

LinkedIn has found that the most common harassing messages it sees are:

• Romance Scams – Members who carry out financial scams through fake or hacked accounts using romantic messaging to defraud a member. 
• Inappropriate Advances – LinkedIn is not a dating website, but some members choose to inappropriately solicit other members for romantic purposes. These members send multiple messages soliciting relationships to members they often don’t know. 
• Targeted Harassment – This includes bringing an off-platform conversation or dispute onto LinkedIn, such as stalking or trolling. These violations are less common and may originate from fake accounts or real members.

​Taking these examples into account, LinkedIn has built an algorithm that can now detect harassing messages based on a three-step process:

1. First, sender behavior (e.g., site usage, invitations sent) is scored by a behavior model. This model is trained using members that were confirmed to have conducted harassment (surfaced via member reports).
2. Second, content from the message is scored by a message model. This model is trained using messages that have been reported as and confirmed to be harassment. 
3. Finally, the interaction between the two members in the conversation (e.g., how often do they respond to one another, are most of the messages predicted to be harassment by the message model) is scored by an interaction model. This model is trained using signals from the conversations resulting in harassment. 

When a potential problem is detected, based on these parameters, LinkedIn will now hide the incoming message, and provide an option for the recipient to either read it or not – or submit a harassment report.

The process will reduce instances of harassing behavior on the platform, and should help to make all LinkedIn members feel safer on the platform. 

Again, any type of harassment is not welcome, but on LinkedIn, where people are working to build professional connections, it seems overly inappropriate. If you’re thinking of sending a romantic proposal via InMail, don’t. Just don’t do it. That’s not what people are on LinkedIn for.

You can read more about LinkedIn’s new harassment detection process here.