LinkedIn introduces a smart reply feature for fast messaging

0
713
smart reply feature for fast messaging

LinkedIn

LinkedIn hopes to boost engagement among its 500 million users with a new smart reply feature for LinkedIn Messaging.

The new feature relies on machine learning to offer quick replies that are relevant to the context of a conversation. According to LinkedIn, the new feature will show up to three responses based on the message a user has received from a contact.

LinkedIn engineers also working on a more personalized smart reply, such as “Thanks, Joe” versus just “Thanks.”

The feature rolling out worldwide, and available only in English on the mobile app and the website for desktops. Users can opt out of the feature in settings. LinkedIn’s help page notes the roll out is happening gradually, so it might not yet be available in all regions.

Unlike Microsoft’s Skype, which uses Cortana for its smart replies, LinkedIn’s engineers developed its system from scratch.

LinkedIn’s explains that, the added controls doesn’t suggest swearing words and doesn’t generate suggestions when user swears in a message.

LinkedIn smart reply feature

To build the feature, the company removes a massive set of real conversations to identify groups of potential synthesized replies while the machine learning message classification model trained on a “very large collection of conversations”.

Conversations automatically scanned by software to find replies corresponding to one of the previously-synthesized candidate replies, explained Nimesh Chakravarthi, a software engineer from LinkedIn’s product engineering team.

The LinkedIn smart reply feature relies on LinkedIn’s instant messaging platform. According to LinkedIn, recipients of a message should see a suggested reply a few milliseconds after receiving the message.

Microsoft completed its $26 billion acquisition of LinkedIn last year and reported LinkedIn revenues were approximately $1.1bn in its Q4 2017 period.

LinkedIn maintained its design earlier this year to focus more on messaging and the Newsfeed, which moved it closer to the look and feel of Facebook.

More information: [LinkedIn]