Expo Workshop

Understanding the user-generated content (UGC) is one of the key components for various applications of machine learning such as social networks, online image/video-sharing platforms, and ecommerce. Today, UGC has evolved from plain texts to more enriched formats including live comments , images, and short videos, which brings new challenges to understand the unique characteristics of these different formats of UGC. Meanwhile, with the development of deep learning techniques, nowadays, we have powerful tools to learn representations of various UGC and extract useful information from them, which enables a wide range of new applications that would otherwise be infeasible due to difficulty in processing such UGC. In particular, this workshop would like to promote research and discussions on the following challenges in the new generation of UGC.

- Heterogeneity (multi-modality) of contents.

- Interconnected entities.

- Large volumes.

- Privacy and ethics issues.

- Data management and analytical systems for UGC.

This workshop will provide a forum for both academic and industrial researchers to review the recent progress of new-generation user-generated content understanding and utilization, with an emphasis on novel approaches and systems that tackle the above challenges. Please find a more detailed workshop description

Chat is not available.


Timezone: »


Sun 4:20 p.m. - 4:30 p.m.
Welcome and Introduction by Workshop Organizers (Talk)
Yaliang Li
Sun 4:30 p.m. - 5:00 p.m.
Broad Learning: A New Perspective on Mining Big Data (Talk)
Philip S Yu
Sun 5:00 p.m. - 5:30 p.m.
User-generated content understanding and its applications in e-commerce (Talk)
Hongbo Deng
Sun 5:30 p.m. - 6:00 p.m.
Query Efficient Adversarial Attacks to User Generated Content (Talk)
Sun 6:30 p.m. - 7:00 p.m.
Multimodal Sentiment Classification and NER (Talk)
Jing Jiang
Sun 7:00 p.m. - 7:30 p.m.
Mediaflow: Orchestration system for streaming media processing and machine learning services (Talk)
Sun 7:30 p.m. - 8:00 p.m.
Collecting Sensitive User-Generated Data Privately (Talk)
Bolin Ding