Topic: Relevance or Profits? Cost-Aware Recommender System Design for Streaming Services
Speaker: Prof. Zhengrui Jiang
Location: Room 306, Glorious Sun Building, Yan'an Road Campus
Time: 2020-12-24 10:00:00
Brief introduction of the speaker: Zhengrui Jiang is a professor, second class professor, and doctoral supervisor in the Department of Marketing and E-Commerce, School of Business, Nanjing University. Prior to joining NJU in 2019, he was Professor of Information Systems and Business Analytics and Tommie Chair Professor at Iowa State University, USA. His main research area is business intelligence and big data analytics, and his research feature is effective integration of computer science research with management research. He has made significant contributions in business data analytics, machine learning, decision support, technology innovation diffusion, etc., and has published more than ten papers in top international journals such as Management Science, MIS Quarterly, Information Systems Research, IEEE Transactions on Knowledge and Data Engineering, etc. His research results have also been applied to enterprises, non-profit organizations and government practices, achieving objective economic benefits. He is currently an Associate Editor of Information Systems Research and Senior Editor-in-Chief of Production and Operations Management, two leading international journals, and was an Associate Editor of MIS Quarterly, another top journal, where he received the journal's 2016 Best Associate Editor Award. He has chaired several international academic conferences in the field of information systems. He has received grants from the National Natural Science Foundation of China and the United States Agency for International Development to conduct research and disseminate knowledge in North America, China and Africa, and was awarded the title of Nanjing High-Level Recommended Talent (Class A) by Nanjing Municipal Government in 2019.
Report Overview: This study proposes a cost-aware recommender system design that balances the relevance and cost of recommendations. The new design uses a control named cost-regularization effort to adjust the weight of cost in relation to relevance when recommending items to consumers. We investigate the cost-aware recommender system design in the context of digital streaming services. Our analytical results show that a streaming platform’s optimal cost-regularization effort increases with a subscriber’s streaming frequency and decreases with the subscription fee, implying that it is beneficial to recommend less relevant but less expensive contents to high-frequency subscribers or subscribers who pay a lower fee. Under the optimal cost regularization effort, when the maximum average utility per session is small, a subscriber’s derived utility decreases monotonically as the subscription fee increases; when the utility per session is sufficiently large, a subscriber’s derived utility first increases and then decreases as the subscription fee increases. We find that, under a discriminatory pricing policy, the optimal subscription fee charged to high-frequency subscribers should be higher than that to low-frequency subscribers, but the same cost-regularization effort should be applied to both subscriber segments. Compared to uniform pricing, discriminatory pricing improves the platform’s profit, and increases the surplus of at least one segments, but possibly both segments, of subscribers. These insights can help digital streaming platforms strategically personalize their recommendations to consumers to achieve a better long-term performance.