One Bite pizza reviews: Building community trust with mobile machine learning

Automated real-time moderation for user-generated content

People are passionate about pizza. Deep dish, thin crust, traditional pan, flatbread — and those are just styles of crust, to say nothing of topping preferences. Pineapple? Anchovies??

With such a wide variety of styles and preferences to swear by, it’s important for pizza lovers to have an authoritative source for honest and legitimate reviews — and the chance to share their love of pizza with others.

One Bite, an iOS and Android app developed by the team at Barstool Sports, delivers just this experience to its users. What started as a popular video review series produced by Dave “El Presidente” Portnoy has evolved into a full-blown, community-driven pizza review platform.

Dave is still doing his thing all around the country, but the true power of One Bite lies in its community of users, who are able to submit their own pizza reviews from more than 100,000 restaurants around the U.S.

This community includes a range of users, from celebrities to everyday people, and their contributions power what has become an authoritative source on the nation’s pizza landscape. But with such a robust community comes some specific challenges.

Verifying user-generated content for a trustworthy community

Specifically, as the One Bite community grew, the problem of image and video content moderation became unmanageable by relying solely on human reviewers.

“Allowing completely open user-generated-content becomes challenging when it comes to verifying submitted content is what users say it is,” said developer Andrew Barba. “In our case pizza, and not other random food or explicit images.”

So Andrew and his team realized that to manage a growing library of user-generated content, they’d have to find a way to automatically do some of this work. For the kind of real-time verification One Bite needed, embedding an ML model in the app became an obvious solution.

“With on-device machine learning, we can tag the recorded clips with the things we recognize in the video,” Andrew noted. “This allows our internal team to much more effectively police inappropriate content.”

Easy model integration and saved resources with on-device machine learning

Given this difficult technical challenge, Andrew and his team turned to Fritz to help implement and maintain One Bite’s ML-powered content review system.

This fast and accurate verification system is a foundational part of the app’s trustworthiness, and thus, the community’s growth and well-being. The app even notifies users when pizza has been identified and verified, which adds more trust and feedback to the experience.

Andrew noted that using Fritz helped One Bite get things up and running quickly while also saving valuable resources in the process.

“Fritz allowed us to quickly integrate real-time image classification while the user is recording the video,” he said. “No data is ever sent to our servers until the user submits their review. This saves us a ton of backend resources and makes for a much faster on-device experience.”

In an age where user-generated content is the lifeblood of so many online communities, finding efficient and accurate ways to moderate that content is an incredibly high priority.

While human review is still an important part of this process, using technology like on-device machine learning to automate initial screening will save time and resources and cement a community’s reputation with its members.

Avatar photo


Our team has been at the forefront of Artificial Intelligence and Machine Learning research for more than 15 years and we're using our collective intelligence to help others learn, understand and grow using these new technologies in ethical and sustainable ways.

Comments 0 Responses

Leave a Reply

Your email address will not be published. Required fields are marked *