Multimodal Event Detection on Social Media

This book systematically explores how to build reliable, cross-platform multimodal event detection systems that operate under open-world conditions in today's complex social media environment. It delves into three core challenges: information fragmentation caused by incomplete individual posts; cross-platform heterogeneity arising from the diverse ways different platforms (e.g., Twitter, Flickr) represent events; and the challenge of open-world discovery driven by the constant emergence of new event types. This book not only reveals how these challenges interact to constrain traditional methods but also provides specialized, cutting-edge solutions for each.

To address these challenges, this book proposes a progressive methodological framework consisting of three specialized models (MFEK, SSMC, and DAEO). By integrating external knowledge, employing self-supervised learning, and using uncertainty-aware new category discovery mechanisms, this framework offers a comprehensive guide from theory to practice. Furthermore, the book contributes several purpose-built datasets (SED, CSED, and MSED) designed to simulate real-world scenarios, aiming to foster reproducible research and bridge the gap between laboratory results and practical deployment. This work serves as a key reference for researchers and practitioners seeking to understand and build the next generation of intelligent event monitoring systems.

This book targets researchers, graduate students  in the fields of Artificial Intelligence, Natural Language Processing (NLP), Multimedia Computing, and Data Mining, who are focusing on multimodal analysis and social media analytics.  Data scientists, software engineers, and industry practitioners working on crisis management systems, public opinion monitoring, misinformation detection, and real-time news aggregation services will also find this book to be a valuable resource.

Zehang Lin received the Bachelor's degree from Guangdong Polytechnic Normal University (Guangdong, China), the Master's degree from Guangdong University of Technology (Guangdong, China), and the PhD degree from the Hong Kong Polytechnic University (Hong Kong SAR). He is currently a lecturer at Hanshan Normal University. His research work has been published in professional journals and conferences such as IEEE Transactions on Multimedia, ACM Multimedia (ACM MM), and Information Processing & Management (IPM). He has served as a reviewer for several top conferences and journals, including AAAI, ACM MM, ICME, ICMR, IPM, and IEEE TAI. His research interests include event detection, cross-domain retrieval, unsupervised learning, and deep learning applications.

Qing Li received his BEng. degree from Hunan University (Changsha, China), MSc and PhD degrees from the University of Southern California (Los Angeles, USA), all in computer science. He is currently a Chair Professor at the Hong Kong Polytechnic University, a visiting professor of the Zhejiang University, a guest professor of the University of Science and Technology of China, and an adjunct professor of the Hunan University. His research interests include multi-modal data modeling, multimedia retrieval and management, and e-learning systems. Dr. Li has published over 500 papers in technical journals and international conferences in these areas, and is actively involved in the research community by serving as a journal reviewer, programme committee chair/co-chair, and as an organizer/co-organizer of several international conferences. Currently he serves as the Chairman of the Hong Kong Web Society, a councillor of the Database Society of Chinese Computer Federation, and a Steering Committee member of the international WISE Society. He is a Fellow of IEEE, AAIA, and IET.