Sequential recommendation via modeling basket sequences

Time: 14:00 to  16:30 Ngày 19/09/2019

Venue/Location: B4-705, VIASM

Speaker: Lê Đức Trọng


The notion of basket-level adoptions is observed in various real-life scenarios. For examples, people purchase a set of items within an online shopping session, or book a package of Point-of-Interests (POIs) for a trip, or assign a set of tags to a bookmark url, etc. Taking into account the time-sensitive manner, a user creates a basket-level adoption sequence from his active sessions. There might exist basket-oriented associations among basket items and sequential associations across sequence baskets. This talk presents two research works on exploiting the two association types concurrently for sequential recommendation task. The first work investigates heterogeneous and contemporaneous trail of actions (e.g., clicks, bookmarks, purchases). Given a sequence of a particular type (e.g., purchases) – referred to as the target sequence, the next item prediction may be helped by also learning from another contemporaneous sequence (e.g., clicks), referred as the supporting sequence. In the latter work, the correlations among basket items are leveraged to enhance the representation of baskets, and subsequently generate correlation-sensitive next-basket recommendations.  

Affiliation: Data Science & Knowledge Technology Laboratory, Faculty of Information Technology, University of Engineering and Technology, Vietnam National Unversity, Hanoi (VNU-UET). 

Short Bio: Le Duc Trong received his bachelor degree in Information Technology from University of Engineering and Technology, Vietnam National University, Hanoi (2011); PhD degree in Information Systems from Singapore Management Univeristy (2019). His research works are published in the proceedings of top-tier confereces on machine learning (ECML PKDD) and artificial intelligent (IJCAI). Recently, he has joined Data Science & Knowledge  Technology Laboratory, VNU-UET  as a research scientist. His research focuses on recommender systems, computer vision and web mining.