HKUST PhD Student Conducts Big Data Analytics on User-Shared Images on Social Media and Achieves Breakthroughs in Smart Living Recommendations
Mr Ming CHEUNG, a PhD candidate at the Hong Kong University of Science and Technology (HKUST), has analyzed 11 million images shared on social media from over 150 countries/locations, and made use of “social signals” to make smart recommendations on daily living that have never been possible.
An application for patent on this technology has been filed.
“My patent-filed technology is built from a huge number of user-shared images, to which we attach machine-generated labels. On the basis of these labels, we create user profiles and discover user connections using the similarity among the profiles,” said Ming.
“This is different from the existing technology of ‘object recognition’, which merely attempts to recognize objects without making further sense out of the images. Meanwhile, our technology can detect social signals from the images for more accurate connection discovery, which enables us to make more personalized recommendations to the users on different aspects of life,” he said.
“For instance, if we see that a user has uploaded photos on Korean celebrities, Korean architecture and Korean food, we believe that he loves Korean culture, and we will recommend to him Korean cuisine when he searches for a restaurant,” he added. “Meanwhile, if another user has shared pictures related to Hong Kong, our technology will recommend to him a dim sum restaurant instead.”
“Apart from the analytics for ‘food selfies’ images, we also carry out analytics for user-shared images about clothing – if we recognize that a user has shared a lot of sports photos, we will recommend sports gear and equipment to him/her. And if we see that another user likes to upload tourist attractions photos, we will recommend to him/her travel information such as hotel and flight details,” Ming said.
Ming Cheung is a “product” of HKUST – having completed his undergraduate, MPhil and PhD studies all at HKUST. He has published eight papers in world top academic journals, and produced 17 conference papers, two of which were elected “best papers”.
On top of his filed patent in this area, he has been granted five innovation funds or awards.
Ming’s PhD supervisor, Prof James SHE, Assistant Professor of the Department of Electronic and Computer Engineering and Director of the HKUST-NIE Social Media Lab, said, “Social media platforms are growing at a phenomenal pace, and there are limitless possibilities in creating new applications in this area.”
