Outstanding Paper Award at the Korea Software Congress 2023 (KSC 2023)

- Machine Learning Systems Laboratory -

The paper 'A Transformer-based Recommender System Using User Social Information' by Kim Cheol-hee, graduate student in the Department of Computer Science & Engineering (supervisor: Moon Eui-hyun), received the Outstanding Paper Award in the artificial intelligence field at the Korea Software Congress 2023 (KSC 2023), held over three days from 20 to 22 December.

▲ From left: Kim Cheol-hee (master's, 4th semester), Yoon Bo-kyung (master's, 4th semester), Cho Hyun-bin (combined master's–doctoral, 1st semester), Moon Eui-hyun (supervisor)

▲ Certificate of the Outstanding Paper Award, Korea Software Congress 2023 (KSC 2023)

In an environment where a wide variety of items is now being created, sophisticated recommender systems are required to meet users' demand for personalized content consumption. Social information representing relationships between users therefore needs to be exploited alongside the user–item interaction information used in existing recommender systems.

The paper proposes a transformer-based model that uses self-attention to learn the relationships appearing in social information and cross-attention to learn the relationships appearing in interaction information. Experiments on two datasets comparing the proposed model with a deep learning model that does not use social information and a graph neural network model that does show that the proposed model improves RMSE by up to 10.1% and 8.0% respectively.

The Korea Software Congress 2023 (KSC 2023) is an information science and technology conference at which the latest research results across all fields of information technology — artificial intelligence, computer systems, databases and more — are presented.