Abstract:Which pieces of content should only be available with a paid subscription? A lot of Publishers use this business model, putting a “paywall” in front of
Which pieces of content should only be available with a paid subscription? A lot of Publishers use this business model, putting a “paywall” in front of certain content, and they rely on human judgment to decide what to paywall. If the content would have a lot of readers/viewers, it could generate a lot of advertising revenue for a publisher, but much much less if only subscribers could read it. On the other hand, certain content is very effective at getting people to pay for a subscription, if they have to in order to read/view. Predicting these potential outcomes in repeated and rapid decision making scenarios is challenging for human operators. We show that advanced neural network architectures can be used to make such decisions more rapidly and effectively than human operators. Our studies show that these models are able to tease out content that can generate subscriptions while minimizing costs to advertising revenue. This Sophi Content Paywall algorithm have greatly accelerated the number of paying subscribers at The Globe, and is now be deployed for Sophi customers on four continents, helping journalism be more sustainable around the world. In addition, we show that it is possible to train models that personalize paywall decisions in real-time. These models are able to further reduce the cost to advertising revenue while outperforming non-personalized models in terms of getting visitors to subscribe or register.
What You’ll Learn:
How NLP and ML can be applied to solve real-world problems.
Siavash Kazemian is a Manager of Data Science at Sophi where he works with a team of talented Data Scientists to solve AI problems in the media industry.
Siavash received a PhD from the Department of Computer Science at the University of Toronto where he studied ways in which NLP technologies, such as Speech Recognition and Sentiment Analysis, can be used to help users perform real-world tasks. His work appears in a number of peer reviewed publications including the ACL. Siavash has also worked in different capacities for startups in various stages of their journey.
(Wednesday) 10:05 AM - 10:50 AM