​From Single-Target to Dual-Target in Cross-Domain Recommender Systems

2024年9月2日；14:00-15:30

信管学院308会议室

Dr. Yan Wang is currently a Full Professor in the School of Computing, Macquarie University, Australia. He received his PhD from Harbin Institute of Technology (HIT), P. R. China. Prior to joining Macquarie University in 2003, he worked as a Postdoctoral Fellow/Research Fellow in the Department of Computer Science, School of Computing, National University of Singapore (NUS). He has published a number of research papers in international conferences including AAAI, AAMAS, ICDE, IJCAI, NeurIPS, SIGIR, WWW, and journals including CSUR, TIST, TKDE, TKDD, TSC and TWEB. His research interests cover trust management, recommender systems, social computing and service computing.   

Prof. Yan Wang has served on the editorial board of several international journals, including IEEE Transactions on Services Computing (TSC), Service-Oriented Computing & Applications (SOCA) by Springer, and Human-centric Computing and Information Sciences (HCIS) by Springer. He also served as a General co-Chair of IEEE ATC2013, IEEE ATC2014, IEEE MS2015, IEEE ICWS2016 and IEEE CLOUD2017, a Program co-Chair of IEEE SCC2011, ATC2011, IEEE MS2014, IEEE SCC2018, IEEE SOSE2018 and IEEE SCC2019, and a Local Organisation Chair of IEEE DSAA2020.

Prof. Yan Wang's research team has received a number of awards, including three Best (Student) Paper Awards from IEEE SCC2010, IEEE TrustCom2012 and IEEE ICWS2016 respectively, Vice-Chancellor's Commendation for Academic Excellence in PhD thesis for 3 times (2017, 2020, 2020), and the nomination for Australasian Distinguished Doctoral Dissertation for 3 times. Prof. Wang received 2017 IEEE TC-TVSC Outstanding Service Award from the IEEE Technical Committee on Services Computing (TC-SVC), IEEE Computer Society.

From Single-Target to Dual-Target in Cross-Domain Recommender Systems

Recommender System aims to predict to what extent a user may like an item based on historical data. Data sparsity is a classic problem in recommender systems. A solution for it is to capture user preferences by doing cross-domain knowledge transfer. This talk will first introduce a few research studies on cross-domain recommender systems (CDR). The first one proposes a novel mechanism to better obtain user features from the data-richer domain and transfer them to the data-sparser domain for improving recommendation accuracy in the context of single-target recommender systems. The second study proposes the first framework in the literature for dual-target recommendation systems and provides a solution for it, where user features can be obtained from both data-richer domain and data-sparser domain for bi-directional cross-domain transfer. Namely both data-richer domain and data-sparser domain are source domains and target domains, and recommender performance can be improved on both domains simultaneously. Furthermore, a unified framework for multi-target cross-domain recommender systems, and cross-domain recommender and cross-system recommender will be introduced. In the end, some future directions in this area will be discussed.