A Human-ML Collaboration Framework for Improving Video Content Reviews

TL;DR

This paper introduces a human-ML collaboration framework that enhances video content moderation by providing model-generated hints to human raters, improving decision quality and efficiency in identifying policy violations.

Contribution

A novel human-ML collaboration approach that uses human feedback to iteratively improve video moderation models and assist raters with targeted hints.

Findings

5-8% AUC improvement in hint generation models

Enhanced moderation decision quality

Increased annotation granularity within same review duration

Abstract

We deal with the problem of localized in-video taxonomic human annotation in the video content moderation domain, where the goal is to identify video segments that violate granular policies, e.g., community guidelines on an online video platform. High quality human labeling is critical for enforcement in content moderation. This is challenging due to the problem of information overload - raters need to apply a large taxonomy of granular policy violations with ambiguous definitions, within a limited review duration to relatively long videos. Our key contribution is a novel human-machine learning (ML) collaboration framework aimed at maximizing the quality and efficiency of human decisions in this setting - human labels are used to train segment-level models, the predictions of which are displayed as "hints" to human raters, indicating probable regions of the video with specific policy violations. The human verified/corrected segment labels can help refine the model further, hence creating a human-ML positive feedback loop. Experiments show improved human video moderation decision quality, and efficiency through more granular annotations submitted within a similar review duration, which enable a 5-8% AUC improvement in the hint generation models.