Professor Haofen Wang
Title: Semantic Framework for Next-Generation Industrial Knowledge Graphs
Abstract:
In the process of digitalization of enterprises, massive amounts of data have been accumulated. Enterprises need to continuously create value for users, while achieving efficient business management and risk control. This puts high demands on the digital infrastructure of enterprises, and also provides diverse application scenarios for AI technologies such as Knowledge Graph (KG) and Large Language Model (LLM). This talk takes merchant management and risk control as examples to introduce the application of KGs in enterprise digitization. In particular, it emphasizes the requirements for deep context awareness due to the portrait coverage and risk insight of thin data customer groups such as small and medium merchants, new users, and sleeping users. Enterprise-level knowledge management is realizing the transition from binary static to multi-dynamic model. Combined with current industrial applications and research progress, we summarized the possible applications of LLM and KG in enterprise digitalization. We point out that LLM has limited application due to the hallucination problem, while KG has wide applications in reasoning, mining, clue insights, analytical querying, etc. because of its expressive ability, interpretability and high computational efficiency of structured knowledge. At the same time, the dual drive of LLM and KG has huge space due to their complementary capabilities, and it is also predicted to be the key path for the industrialization of LLM in language understanding and interactive applications. On this basis, we further introduce the current challenges of KG technology. Finally, combined with the practice of Ant Group's KG platform, we introduced the industrial-level semantic-enhanced programmable Graph SPG and KG engine co-built by OpenKG.
Short Bio:
Haofen Wang is a professor at College of Design & Innovation, Tongji University. Prior to that, He served as CTOs for two well-known AI startups (i.e., Leyan and Gowild). He is also one of the co-founders of OpenKG, the world-largest Chinese open knowledge graph community. He has taken charge of several national AI projects and published more than 100 related papers on top-tier conferences and journals. He developed the first interactive emotional virtual idol in the world. The intelligent assistant he built has answered questions from more than one billion users when they did online shopping. He has also served as deputy directors or chairs for several NGOs like CCF, CIPS and SCS.
Professor Jiawei Jiang
Title: Graph Learning for Fraud Detection
Abstract:
Fraud detection has become one of the most prominent research topics for e-commerce companies. In this tutorial, I will summarize the efforts we have made to design explainable fraud detection systems using graph neural networks (GNNs). These systems are designed according to various deployment requirements in the fraud detection ecosystem. We have also reviewed our efforts on deploying GNNs in a dynamic graph setting and a real-time fraud detection environment. Through this tutorial, we aim to provide a holistic understanding of the current state and future prospects of our efforts in various stages of fraud detection.
Short Bio:
Jiawei Jiang is a professor in School of Computer Science, Wuhan University. He obtained his B.Sc. from University of Science and Technology of China in 2011, and Ph.D from Peking University in 2018, respectively. From 2019 to 2022, he worked as a postdoc researcher at ETH Zürich. His research interests include, but are not limited to, machine learning systems, large-scale data analytics, graph processing, and federated learning. He has published more than 40 papers in top venues, e.g., SIGMOD, VLDB, ICDE, ICML, and NeurIPS. He has served in the Technical Program Committee of various international conferences including VLDB, ICDE and KDD. He was awarded CCF Outstanding Doctoral Dissertation Award (2019) and ACM SIGMOD Doctoral Dissertation Award (2018).
Professor Jiang Xiao
Title: Blockchain Storage Scalability: State of the Art and Challenges
Abstract:
The realization of trustworthiness Web 3.0 era requires accessing the unprecedented benefits -- auditability, transparency, automation, effectiveness -- from the disruptive blockchain technology. However, as the amount of data is exploding at unprecedented scale, blockchain systems existing today are becoming inefficient in storing and processing the enormous data. In this talk, we will uncover many unexplored challenging issues of blockchain storage scalability. We will also discuss about the initial studies leaving tremendous potential for further innovation.
Short Bio:
Jiang Xiao, currently a full professor at Huazhong University of Science and Technology. Her research interests include blockchain, distributed systems, etc. She has hosted the National Key R&D Program Young Scientist Project, National Natural Science Foundation Project, Hubei Provincial Key R&D Program Project, etc. She published more than 70 papers in international journals and academic conferences such as IEEE TPDS, IEEE TKDE, IEEE JSAC, INFOCOM, SRDS, ICDCS, and ACSAC. She served on the editorial board of several international blockchain academic journals and as the chairman of several international blockchain academic seminars. Her awards include CCF-Intel Young Faculty Research Program 2017, Hubei Downlight Program 2018, ACM Wuhan Rising Star Award 2019, Knowledge Innovation Program of Wuhan-Shuguang 2022, and Best Paper Awards from IEEE ICPADS/GLOBECOM/ BLOCKCHAIN /GPC.