Trustworthy Recommender Systems

Trustworthy Recommender Systems











Recommender systems havealready become an indispensable means for helping web users identify the mostrelevant information/services in the era of information overload. Theapplications of recommender systems are multi-faceted, including targetedadvertising, intelligent medical assistant, and e-commerce, and are bringingimmense convenience to people’s daily lives. However, despite rapid advances inrecommender systems, the increasing public awareness of the trustworthiness of relevantweb applications has introduced higher expectations on relevant research.Firstly, the unprecedentedly growing heterogeneity of use cases has beenchallenging the adaptivity of recommender systems to various settings, e.g., dynamicinterest drift of users, cold-start users/items, and highly sparse interactionrecords in large-scale datasets. Secondly, in a broader sense, a trustworthy recommendationapproach should also be robust, interpretable, secure, privacy-preserving, andfair across different use cases. Specifically, robustness evaluates a model’sperformance consistency under various operating conditions; interpretabilityand fairness respectively evaluate if a model can make its decision processestransparent and the decision outcomes unbiased; while security and privacy respectivelyemphasize a model’s ability to handle cyber-attacks and to prevent personalinformation breaches. Consequently, the trustworthiness is becoming a keyperformance indicator for contemporary recommender systems in addition toaccuracy. In light of these emerging challenges that co-exist with existingrecommendation techniques and applications, this special session focuses onnovel research on recommender systems with the notion of trustworthiness. Thedevelopment of trustworthy recommender systems will further promotehuman-in-the-loop AI applications, thus better universalizing the advancedtechniques to a wider range of the common public.  


Hongzhi Yin, The University of Queensland, Australia.


Hongzhi Yin works as an associate professor with TheUniversity of Queensland, Australia. He was recognized as Field Leader of DataMining & Analysis in The Australian’s Research 2020 magazine. He receivedhis doctoral degree from Peking University in July 2014, and his PhD Thesis wonthe highly competitive Distinguished Doctor Degree Thesis Award of PekingUniversity. His current main research interests include recommender systems,graph embedding and mining, chatbots, social media analytics and mining, edgemachine learning, trustworthy machine learning, decentralized and federatedlearning, and smart healthcare. He has published 160+ papers, including 13publications in Top 1% (CNCI), 90+ CORE A* and 50+ CORE A. He is the leadingauthor (first author or corresponding author) for 90+ of them. He has won 6Best Paper Awards such as ICDE’19 Best Paper Award, DASFAA’20 Best StudentPaper Award, and 21st ACM Annual Best of Computing Article as the first author.He has received ARC Discovery Early Career Researcher Award (DECRA) within hisfirst year of obtaining his PhD, ARC Discovery Project 2019 (Sole CI) as anearly-career researcher, UQ Foundation Research Excellence Award 2019 as thefirst winner of this award in School of ITEE since the establishment of thisaward 20 years ago. He is currently directing the Responsible Big DataIntelligence Lab (RBDI). RBDI Lab aims and strives to develop decentralized,on-device, and trustworthy (e.g., privacy-preserving, robust, explainable andfair) data mining and machine learning techniques with theoretical backbones tobetter discover actionable patterns and intelligence from large-scale,heterogeneous, networked, dynamic and sparse data. RBDI joins forces with otherfields such as urban transportation, healthcare, agriculture, E-commerce andmarketing to help solve societal, environmental and economical challengesfacing humanity, in pursuit of a sustainable future. His research has also beenattracting media coverage, such as UQ News, Computing Reviews, and 360 News.

Tong Chen, The University of Queensland, Australia.


Tong Chen is a postdoctoral research fellow with the Data Science Discipline at TheUniversity of Queensland. Before that, he received his PhD degree in ComputerScience from The University of Queensland in 2020. His research interestsinclude data mining, machine learning, recommender systems, and predictiveanalytics. He has 35+ publications on top-tier (CORE A*/A) international data mining/machinelearning venues such as KDD, SIGIR, ICDE, AAAI, IJCAI, ICDM, WWW, TKDE, IJCAI,TOIS, CIKM, as well as the prestige health informatics journal JBHI (CORE A*).He is the recipient of Best Student Paper awards of DASFAA’20 and PAKDD’18Workshop.  He has been serving as acommittee member or reviewer for over 20 world-leading internationalconferences/journals in the fields of data mining, information, and AI,including IJCAI’21 (Senior PC), CIKM’21 (PC), IJCAI’20 (PC), TKDE (journalreviewer), TOIS (journal reviewer), TNNLS (journal reviewer), etc. Meanwhile, hehas also been a session chair for conferences VLDB’20 (the top-1 Databaseconference, CORE A*), BESC’20, and DASFAA’20. Dr. Chen’s recent research has beenfocusing on trustworthy and lightweight recommender systems to embrace the eraof edge computing (e.g., internet-of-things), which is witnessed by hispublications on KDD’21, WWW’21, SIGIR’20, ICDE’20, and WWW’20, as well as hisinvited talk about interpretable recommendation on the ICDM’20 NeuRec Workshop (,150+ attendees).  

Other Workshops