MAPS: A Multi-Agent Personalized Script-Driven Video Advertising System Enhanced by Reinforcement Learning

Key points

• With the rapid growth of online video consumption, video advertising has become a dominant component of digital advertising.1
• Most personalized video advertising systems follow a retrieval-based paradigm, which matches ad content with user preferences in a feature space and recommends the most similar candidates.1
• To train the Scripter Agent, we develop a reinforcement learning framework equipped with a multi-dimensional reward function, which steers the agent toward producing scripts that simultaneously align with user content preferences and target product information.1
• Online A/B experiments in a specific scenario in the Kuaishou advertising system show that MAPS improves CTR by 2.87%, CVR by 2.74%, and ARPU by 9.85% over the generative baseline.1

Sources (1)

• [1]MAPS: A Multi-Agent Personalized Script-Driven Video Advertising System Enhanced by Reinforcement Learning
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 19, 04:00 AM

  To address this limitation, we propose a multi-agent personalized script-driven video advertising system (MAPS).

  With the rapid growth of online video consumption, video advertising has become a dominant component of digital advertising.

Extractive summary: sentences quoted from the sources.