Artificial Intelligence is becoming more prevalent in our services, products, and our lives. The popular online news and social media service, Twitter is now using deep learning, an Artificial Intelligence (AI,) tool to sort and recommend tweets to a user regarding the timeline.
Twitter has 328 million monthly active users. To serve that huge user base, it is now using AI to sort and arrange tweets on the user’s timeline by replacing an algorithm used previously. Twitter used decision trees, logistic regression, feature crossing and discretization algorithms to rank tweets on users timeline. Now with this deep learning algorithm, they will be ranking the tweets by relevance rather than in reverse chronological order.
Nicolas Koumchatzky and Anton Andryeyev, Twitter software engineers, posted on their blog that the new algorithm is powered by deep neural networks, this is developed by one of the in-house Artificial Intelligence teams called Cortex.
Magic Pony Technology, a company which developed the artificial intelligence and machine learning based approaches for mobile, desktop and web, was recently acquired by Twitter and is now helping with Artificial Intelligence research for the company.
Twitter is also considering likes, retweets, photos and videos to determine the order of relevant tweets. As the statistics state user engagement is increasing on Twitter. So, the research and development team is working to enhance this user experience.