Advances in Simulation Technology for Web Applications (Information Retrieval, Recommender System, Advertising)
(The Web Conference 2023 Tutorial, Austin, Texas)
The research and development of modern Web applications, including search, recommendation, and advertisement, are highly complex and iterative processes that require a deep understanding of system dynamics and user behavior. Simulations have proven useful in tackling various unresolved issues in both academia and industry, such as measuring the effect and potential bias of system interventions, forecasting long-term benefits such as equity, inclusivity, and diversity, and assessing intricate system characteristics under diverse human behaviors. Although simulations enable researchers and developers to craft controlled environments, there is a fragile equilibrium between manipulability and faithfulness, flexibility and complexity, hypothesis and actuality, which can raise doubts regarding the validity and practicality of simulation analysis.
Indeed, the efficacy of Web simulations is heavily reliant on the extent of analysis, the underlying assumptions, and the specific objectives and tasks at hand. Although simulations are widely employed by the community, there is a deficiency of comprehensive and structured tutorials that furnish instructions and guidelines for selecting appropriate tools in a given context.
Our tutorial has been proposed in a timely manner, with the objective of providing attendees with the essential background and scope, summarizing existing simulation concepts and frameworks, analyzing their main factors and limitations in real-world case studies, and linking Web simulations to advanced technologies such as causal inference, reinforcement learning, and Generative AI. We aim to present a comprehensive overview of the various factors that need to be considered, enabling attendees to navigate through the numerous intricate choices they may encounter while utilizing simulation technologies in practice. Drawing on the presenters' expertise and insights from other works, this lecture-style tutorial is intended to assist in designing, implementing, and optimizing simulation analysis for a variety of Web applications.
Logistics
Tutorial Date and Location:
The tutorial will be held during the afternoon session (3pm - 5pm CDT) on May 1, 2023, at AT&T Hotel and Conference Center at The University of Texas at Austin.
Attendence and Registration:
All onsite attendees must be registered. Please refer to the The Web Conference Registration website for more information regarding the registration. The tutorial will be recorded and uploaded to the conference website as well.
Please contact daxu5180 at gmail.com for questions.
About the Speakers
Da Xu
Staff AI Engineer
Da Xu is a Staff AI Engineer at the Network Growth AI team at LinkedIn. He was previous a Manager of Machine Learning at the Search & Recommendation team of Walmart Labs. After joining the industry from UC Berkeley in 2018, Da has been driving research and realworld productions that push the frontier of modern IR systems. In the past several years, his research works that invent theoretical tools for modern IR systems have been published in major ML/AI conferences, including NeurIPS, ICML, ICLR, AAAI. His industrial and application work has been published in such as KDD, WSDM, WWW. Da is also actively engaged in community public services, he was designated by the INFORMS committee as the session chair of Causal Inference Analysis for Information Retrieval for 2021 and 2022. He organized the WSDM'22 and WWW'23 Workshop on Decision Making for Information Retrieval Analysis, and KDD tutorial on Theoretical Foundation for Information Retrieval and Recommender Systems
Shuyuan Xu
Ph.D Candidate
Rutgers University
Shuyuan is a PhD student in the Department of Computer Science at Rutgers University supervised by Prof. Yongfeng Zhang. His research interest lies in the intersection of Machine Learning and Information Retrieval. His current research focuses on causal inference for recommender systems, including counterfactual explainable recommendation, deconfounded recommendation, causal structure learning for recommendation. He is actively serving as reviewer for conferences or journals such as AAAI, WWW, CIKM, ACM TOIS, ACM TORS, IEEE TKDE.
Bo Yang
Applied Scientist
Amazon
Bo has been a Machine Learning Engineer at LinkedIn’s Ads AI since 2021. She is responsible for developing the auction, bidding, and decision-making strategies for LinkedIn marketplace systems. She obtained Ph.D in Statistics from University of Virginia in 2019, with emphasis on causal inference and time sires. Bo first joined Target AI as a lead ML scientist in 2019, and she has been serving as PC members for such as KDD, SIGIR, WWW, and co-organizing and presenting at workshops and tutorials for such as WSDM, KDD, and INFORMS.
Yongfeng Zhang
Assistant Professor
Rutgers University
Yongfeng is an Assistant Professor in the Department of Computer Science at Rutgers University (The State University of New Jersey). His research interest is in Machine Learning, Machine Reasoning, Information Retrieval, Recommender Systems, Explainable AI, Fairness in AI, and AI Economics. In the previous he was a postdoc at UMass Amherst, and did his PhD and BE in Computer Science at Tsinghua University, with a BS in Economics at Peking University. He is a Siebel Scholar of the class 2015 and an NSF Career Awardee in 2021. Together with coauthors, he has been consistently working on explainable search and recommendation, fairness-aware recommendation, echo chambers in IR systems, as well as causal/counterfactual models for search and recommendation. He has served as PC members or senior PC members in various Web\&IR related conferences such as SIGIR, WWW, KDD, RecSys, CIKM, WSDM, ICTIR and CHIIR, and he is serving as the associate editor for ACM Transactions on Information Systems (TOIS) as well as ACM Transactions on Recommender Systems (TORS). He has experience in organizing workshops on SIGIR and WSDM. He also has experience organizing tutorials in top conferences such as the Tutorial on Explainable Recommendation and Search presented at WWW'19, SIGIR'19 and ICTIR'19, the Tutorial on Conversational Recommendation presented at RecSys'20, WSDM'21 and IUI'21, as well as the tutorial on Fairness of Machine Learning in Recommender Systems at SIGIR'21 and CIKM'21.