[Mingli Lecture Hall 2022 Issue 27] 

  At the invitation of the School of Management and Economics, Assistant Professor Xie Jiaheng from the Alfred Lerner School of Business, University of Delaware came to our school for an exchange visit, and made an online speech entitled "Unbox Academic report on the Blackbox: Predict and Interpret YouTube Viewership Using Deep Learning. The report meeting was presided over by teacher Yin Qiuju, and many teachers and students of the college attended the report meeting.

  Professor Xie Jiaheng first introduced the relevant background of online video websites to everyone, pointing out that content creators and video platforms are eager to predict the viewership rate of videos and optimize the influence and marketing effect of videos at low cost. Although currently widely used deep learning models can predict video influence with high accuracy, they lack interpretability. However, whether video producers and platforms accept model results and make decisions based on them requires high model interpretability. Against this research background and combined with the design science paradigm, Professor Xie and his collaborators proposed a predictive model called Precise Wide-and-Deep Learning (PrecWD), which utilizes unstructured video information and is extracted from The features accurately predict the video ratings, and at the same time have better interpretability on the impact of the features. PrecWD outperforms baseline models on multiple performance metrics and exhibits excellent interpretability in user survey experiments.

  After the report, Prof. Xie briefly shared another cutting-edge research he is currently conducting, and the participating teachers and students had a positive and full discussion with Prof. Xie. The report received enthusiastic response and was well received by teachers and students.

 Assistant Professor Xie Jiaheng:

  Dr. Xie Jiaheng is an Assistant Professor in the Department of Accounting and Management Information Systems at the Alfred Lerner School of Business at the University of Delaware. He received his Ph.D. from the Eller School of Management at the University of Arizona. His research interests include deep learning, health risk analysis, and business analytics. His previous work has been published in many leading journals, including MIS Quarterly, Journal of Management Information Systems, and Journal of American Medical Informatics Association.