管理科学系学术讲座

时间：2024年5月30日（周四） 10:00-11:30

地点：管理学院思源楼524室

主题：Natural Language Processing Applications in Operations

主讲人：吴迪密西根大学副教授

Associate Professor of Technology and Operations (with tenure) at the Ross School of Business, University of Michigan

主持人：冯天俊复旦大学管理学院教授

摘要:

I present two novel use cases of natural language processing (NLP) models in Operations. In the first study, using a new technique of "seeded word embedding", I develop a firm-level measure of supply chain risk exposure from the audio-recorded discussion between managers and equity analysts on supply chain--related topics. I validate the measure by showing that the measure exhibits intuitive variations over time and across firms, and it significantly correlates with realized and options-implied stock return volatility. I then demonstrate that, consistent with theoretical predictions, firms facing higher supply chain risks have higher inventory buffers, particularly in raw materials and intermediate inputs, increased cash holdings in lieu of investments, and significantly lower trade credit received from suppliers. Moreover, during unexpected risk episodes such as the Tohoku earthquake, firms with higher ex-ante risk exposure have worse operating and financial performance. These results indicate that the text-based measure provides a credible quantification of firm-level exposure to supply chain risks, and can thus be reliably utilized as outcome or explanatory variables in empirical supply chain research.

The second project develops and trains a generative large language model (LLM) as an "AI tutor" in higher education. In collaboration with the University of Michigan's IT team, we developed a GPT-powered AI---"Maizey"---that is trained using individual faculty's course content to provide immediate, accurate responses to student inquiries. The AI automatically integrates into course Canvas sites, then creates a knowledge repository for each course by absorbing all text, audio and video contents from the Canvas course. The model automatically retrieves from the repository the most relevant course content for each student query, then employs GPT's generative capabilities to deliver direct and relevant answers in natural language. With implementation across courses involving thousands of students, we aim to enhance user satisfaction, optimize the utilization of teaching staff time, and improve student learning outcome.

简介：

Andrew Wu is the Associate Professor of Technology and Operations (with tenure) at the Ross School of Business, University of Michigan, and the Faculty Co-Director for the Michigan Ross Fintech Initiative. He is also the Michael R. and Mary Kay Hallman Faculty Fellow. His research centers on unlocking new business insights from human language and other unstructured data. He develops innovative natural language processing (NLP) and computer vision methodologies to extract operational intelligence, strategic insights, and trading signals from diverse sources such as corporate communications, news and social media, investor and executive interviews, etc.

His work received numerous research and practice awards, including the BlackRock Applied Research Award, seven Best Paper Awards from INFORMS, M&SOM, POMS, and finance societies, and the finalist for the Financial Times‘ Research Impact Award. His methodologies have been adopted by a variety of industry practitioners, nonprofit and government organizations, and investment managers. He is recognized as one of Poets & Quants' Best 40 Under 40 Business School Professors.

时间：2024年6月7日（周五） 10:00-11:30

地点：管理学院思源楼524室

主题：Competitive Pricing at Scale: Theory and Practices

主讲人：李君 Professor of Technology and Operations at Stephen M. Ross School of Business, University of Michigan

主持人：冯天俊复旦大学管理学院教授

摘要:

In this talk, I will present a series of theoretical and applied work in competitive pricing. A retailer following a competition-based dynamic-pricing strategy tracks competitors’ price changes and then must decide whether and how to respond. The answers require modeling of consumer decisions, unbiased measures of self- and cross-price elasticity as well as competitor impacts. I will discuss how we achieve them through a combination of consumer modeling, experimentation, causal inference, as well as high-dimensional statistics. I will highlight two implementations that each led to 10-20 percent revenue and profit improvement through close collaborations with leading ecommerce retailers, and a theoretical development that scales up choice models through high dimensional regularization.

简介：

Jun Li is a Professor of Technology and Operations at Stephen M. Ross School of Business, University of Michigan. She conducts research in empirical operations management and business analytics spanning areas across revenue management and pricing, healthcare management, supply chain risks, and public sector operations as well as improving the wellbeing of children and young adults. She won the 2015 INFORMS Revenue Management and Pricing Practice Award, the 2020 INFORMS Revenue Management and Pricing Section Award, the 2022 MSOM Young Scholar Prize, Poet&Quant 40 Under 40 MBA Professors, the Management Science Best Publication award and the Responsible Operations Management Best Publication award, etc. She serves as Associate Editors at Management Science, Manufacturing and Service Operations Management, and Operations Research. She holds a Ph.D. in Managerial Economics and Management Science from the Wharton School, University of Pennsylvania, and a Bachelor in Operations Research and Industrial Engineering from Tsinghua University.