KDD 2022 in Washington, D.C.

In August 2022, Hady, Zhang Ce, and Thuat travelled to Washington D.C. to attend the 28^th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD for short). This year, KDD came back to an in-person event after two years of virtual events. The conference welcomed more than two thousand attendees, including both academic and industrial researchers. 

As a premier forum for presentation of cutting-edge research in data science, KDD is competitive with around 15% acceptance rate for research track over the last three years. In 2022, there were 1,659 submissions for research track, and 254 out of them were accepted, resulting in 15.3% acceptance rate. The first two days of the conference were filled with workshops and tutorials together with three keynote sessions, spanning a wide range of topics. Poster and presentation sessions of research papers were the main theme in the remaining conference days. 

This year, Preferred.AI contributed two research papers to KDD. The first paper titled ‘Variational Author Topic Modeling’, authored by Zhang Ce and Hady, proposes to incorporate semantic interpretability and authorship modeling into a unified VAGE framework to enhance document representation. This paper was selected to conduct a 15-minute oral presentation in Text Mining session. 

Our group’s second paper, authored by Thuat and Hady, titled ‘Aligning Dual Disentangled User Representations from Ratings and Textual Content’, was presented in Recommender Systems session. The paper proposes to learn a set of disentangled user’s representations from textual content and align them with disentangled representations derived from ratings to improve recommendation performance and interpretability. 

We also attended a 90-minute poster session, which engaged more audiences to dive deeper into specific problems. Poster session was also a great chance to learn from other researchers’ experience as well as get to know more people in the field. 

Washington D.C. is a diverse city and attracts tourists with its own unique history, beauty, and events. During our time here, we had unforgettable experiences and made lasting memories.

KDD is always a great venue for high quality and leading research in data science. Preferred.AI is thriving on hard work to present in future KDD!