题    目：Sequencing Visual Concepts in Video Advertising: An Interpretable Storyboard Approach

嘉　宾：Runyang Shu（舒润阳）, Ph.D. Candidate, University of Texas at Austin

主持人：左思 助理教授 上海交通大学安泰经济与管理学院

时　间：2026年11月4日（周三）13:30-15:00

地　点：上海交通大学安泰经济与管理学院包兆龙图书馆A303

内容简介:

Given a set of candidate clips featuring people, products, usage, or branding, how should advertisers decide what comes first, what follows, and what closes the ad? We address this creative decision by developing a framework that learns a shared vocabulary of recurring visual concepts—such as presenter shots, product-use scenes, product information, promotional text, and brand logos—from historical advertising data. Once learned, this shared vocabulary allows the framework to represent heterogeneous advertisements as comparable, interpretable storyboards and use those representations to predict brand recall. Applied to 2,205 real-world advertisements, our modeling framework approaches the predictive performance of black-box video models while providing timing suggestions for creative edits. We then test one such edit in a controlled resequencing experiment. Holding the visual clips fixed, the model-guided ordering increases unaided brand recall relative to both the original ordering and randomly shuffled controls. Together, our framework shows how historical advertising data can be used to study the temporal organization of ad content at scale and generate candidate sequencing revisions for consumer testing.

演讲人简介：

Runyang Shu is a Ph.D. candidate in Quantitative Marketing at the University of Texas at Austin. His work studies how AI can help firms understand consumers, design better digital content, and govern platform experiences. Runyang develops interpretable AI methods for visual and video data, with applications to product recommendation, consumer preference modeling, and advertising effectiveness. He also collaborates with technology companies to run field experiments and study AI adoption and governance in real operational settings. His research interests include AI for marketing, multimodal measurement, visual analytics, and field experimentation. Previously, Runyang worked as a research scientist focused on responsible AI.

欢迎广大师生参加!