Application of Algorithm Strategies in Live-commerce Merchants for Live-streamers Selection: Reducing the Risk of Anomalous Behavior through Network Dynamics and Adaptive Control

In the rapid evolution of the digital economy and e-commerce, live-streaming e-commerce has become a key sales channel. However, its reliance on top-tier live-streamers exposes businesses to significant risks of frequent abnormal behaviors, such as issues related to live-Streamer Crisis Incidents. To address this challenge, this study introduces algorithm-driven strategies to optimize the selection of live-streamers for live-commerce merchants, utilizing complex network dynamics and adaptive control models for risk prevention. The research proposes a two-layer network framework: an improved Hegselmann–Krause (HK) model is applied to the scale-free live-streamers-fan network, while an adapted Deffuant–Weisbuch (DW) model is applied to the small-world fan-fan social network. By simulating opinion evolution in the information flow network of live-streaming e-commerce, the study quantifies the impact of live-streamers abnormal behavior on purchase intentions, demonstrating the effectiveness of the proposed algorithm and its compatibility with live-streaming e-commerce scenarios. This study integrates computer science with live-streaming e-commerce management, showcasing how network dynamics algorithms and adaptive control strategies can systematically reduce dependence on high-risk top-tier live-streamers, providing a data-driven framework for robust and sustainable operations in live-streaming e-commerce.