Chinese Morph Resolution in E-commerce Live Streaming Scenarios

TL;DR

This paper introduces the LiveAMR task to detect pronunciation-based morph evasion in Chinese e-commerce live streams, creating a large dataset and leveraging large language models to improve regulation accuracy.

Contribution

It is the first to address pronunciation-based morph resolution in live streaming, transforming the problem into text-to-text generation and utilizing LLMs for data augmentation.

Findings

Constructed the first LiveAMR dataset with 86,790 samples.

Transform the morph resolution task into a text-to-text generation problem.

Using LLMs improves morph resolution performance.

Abstract

E-commerce live streaming in China, particularly on platforms like Douyin, has become a major sales channel, but hosts often use morphs to evade scrutiny and engage in false advertising. This study introduces the Live Auditory Morph Resolution (LiveAMR) task to detect such violations. Unlike previous morph research focused on text-based evasion in social media and underground industries, LiveAMR targets pronunciation-based evasion in health and medical live streams. We constructed the first LiveAMR dataset with 86,790 samples and developed a method to transform the task into a text-to-text generation problem. By leveraging large language models (LLMs) to generate additional training data, we improved performance and demonstrated that morph resolution significantly enhances live streaming regulation.