AWS for Games Blog

Introducing Community Health on AWS

Amazon Web Services (AWS) is proud to announce the launch of Community Health on AWS, a set of solutions designed to help AWS for Games customers build and grow safe, inclusive, and fun online communities. Initial Community Health on AWS solutions include Spectrum Guardian for Games, a fully managed software-as-a-service (SaaS) offering from AWS Partner Spectrum Labs that runs on AWS; reference architecture for building or modifying content moderation pipelines on AWS; and AWS for Games Cohort Modeler.

Online communities are often essential for successful game development, but protecting them and keeping tabs on inappropriate behavior has become more difficult in recent years. A 2021 study by the anti-defamation league found that 83 percent of adults experienced harassment while playing online games, and 67 percent of respondents to a 2021 Unity Technologies study said they would likely leave a game if they experienced disruptive or toxic behavior.

Community Health on AWS solutions leverage machine learning (ML) to help alleviate this burden for AWS for Games customers, enabling more efficient and effective moderation to protect online communities. Using natural language processing that extracts topic, intent, and sentiment from text and computer vision, which does a similar function for images and video, alongside deep learning, problematic activity can be quickly flagged, and corrective measures implemented.

An end-to-end solution, Spectrum Guardian for Games uses AI natural language processing to detect harmful behaviors in player-generated content like direct messages, in-game chat, and usernames. Major studios, including Riot Games, ensure more positive player experiences by using the turnkey solution to identify and remediate situations with bad actors.

Existing content moderation pipelines that require additional innovation can be augmented using the Community Health on AWS reference architecture. It provides access to fully-managed SaaS AI tools, including Amazon Rekognition for image and video analysis, Amazon Comprehend for text analysis, Amazon Transcribe for voice stream analysis, and Amazon Augmented AI (Amazon A2I) for technology oversight. The reference architecture can also be used to build new pipelines, and AWS Professional Services can be enlisted to create bespoke solutions based on customer needs.

A social behavior mapping solution built on the Amazon Neptune graph database, AWS for Games Cohort Modeler helps identify and track positive and negative behavior across an online community. Its source code is available on GitHub, and details about the solution can be found in this deep dive blog post.

We look forward to helping customers forge thriving communities in collaboration with our partners as we continue building out solutions for Community Health on AWS.

 Resource links: