OpenWeb: Combatting Online Hate Speech and Violence Through Advanced Content Moderation

OpenWeb, a social engagement platform that builds online communities around better conversations, needed a way to scale their expanding data science team’s efforts and show immediate value.

The rise of digital platforms has made communication more accessible than ever before, but it has also paved the way for a surge in online hate speech and violent content. Addressing this growing concern, many companies are turning to innovative technological solutions.

One such solution is the deployment of AI-powered content moderation systems. This case study delves into the collaborative journey of a new data science team and Qwak, a key player in the AI sphere.

Together, they took on the ambitious task of refining and scaling a system capable of detecting and mitigating harmful online content. Through their combined efforts, they aimed not just to keep the digital space safe, but also to showcase the transformative power of technology when applied with precision and purpose.

About OpenWeb

OpenWeb is constructing the healthy, social layer of the internet, partnering with publishers and brands to enhance audience relationships and improve online conversations, fostering an internet where content creators of every kind can independently own and thrive with their audience relationships.

Industry

Internet, Social Networks, Publishing, Spam Detection

Use Case

Content discussion moderation

Model Frameworks

Pytorch

Qwak's service is impeccable and the platform provides immediate feedback which allows us to turn things around quickly when needed.

Idan Benaun
Director of ML and Data Science
<50ms

Real-Time Inference

Centralized

Model Repository

5

New Models Introduce in no time

Challenges

As OpenWeb's data science team expanded, there was a pressing need to demonstrate immediate value to stakeholders. Their challenges included:

  • The team's rapid growth meant they required engineering solutions that enabled them to swiftly move forward without the burdens of server maintenance and infrastructure configurations.
  • Previous solutions tested presented intricate change management and basic configurations that hindered smooth execution.
  • The team had a crucial need for real-time inference responses, necessitating the processing through multiple models in under 50 ms.
  • Instantaneous model deployment was a non-negotiable requirement.

Solutions

As OpenWeb's data science team expanded, there was a pressing need to demonstrate immediate value to stakeholders. Their challenges included:

  • The team's rapid growth meant they required engineering solutions that enabled them to swiftly move forward without the burdens of server maintenance and infrastructure configurations.
  • Previous solutions tested presented intricate change management and basic configurations that hindered smooth execution.
  • The team had a crucial need for real-time inference responses, necessitating the processing through multiple models in under 50 ms.
  • Instantaneous model deployment was a non-negotiable requirement.