Proprietary Natural Language Processing (NLP) approach outperforms accuracy of industry-standard solutions in assessment of user generated content.
Conversation analytics platform Spiketrap today revealed a brand safety suite designed specifically to address the unique challenges presented in accurately classifying high-velocity user-generated content.
The solution, which uses Spiketrap’s proprietary Clair AI technology to power Natural Language Processing (NLP) for the accurate identification of unsafe content, provides clear grading across several categories of concern for advertisers and creators alike, including hateful, sexual, and profane content.
The brand safety suite provides content creators, platforms, and brands with a reliable and independent source of safety verification for any channel or conversation. The solution is accessible both through a dashboard and as an API integration, which enables real-time monitoring.
The rapid growth of the creator economy, both with respect to content production and audience engagement, has amplified the need to verify the safety of highly fragmented, high-velocity digital environments.
“Livestreaming enables content creators to meaningfully engage with their audiences while creating a dynamic environment that offers significant opportunities for brands,” said Sean Horvath, CRO at StreamElements. “In the last year, we experienced more than a 300% increase in adoption of our streaming tools, making Spiketrap’s brand safety offerings very timely.”
As noted by Chris Kindt, CMO at Gaming Community Network, “Knowing the safety of a given chat environment is absolutely critical to both the brands we work with and our media monetization strategies, and Spiketrap is our trusted provider for live stream community brand safety reporting and analysis.”
Spiketrap’s brand safety solution uniquely leverages NLP AI to measure live environments, setting it apart from historical approaches that only measure static source material or that depend on keywords — which are inaccurate and lack needed context — to flag content.
A blind study conducted by the company found that Spiketrap’s Brand Safety API significantly outperforms both the industry-standard toxicity detection API as well as keyword lists relied upon by the Fortune 500. Specifically, in a comparison of F1 scores (an indicator of ability to accurately identify unsafe content without false positives or negatives) Spiketrap’s brand safety solution scored 32% higher than tested keyword lists used by Fortune 500 brands and 11% higher than the industry standard API alternative.
“Spiketrap was founded to facilitate language understanding at scale. Our proprietary Clair AI coupled with our organically-growing knowledge graph uniquely position us to provide the brand safety scoring that has become prerequisite table stakes for the platforms, brands, and communities inspiring authentic audience conversations,” said Kieran Seán Fitzpatrick, Founder and CEO at Spiketrap.