PhD Dissertation Proposal by LE Duc Trong | Modeling Sequential and Basket-Oriented Associations for Top-K Recommendation

About The Talk

Top-K recommendation is a typical task in Recommender Systems. Previous works focus on the user-item associations, which emphasize the user-specific factor or personalization. Here, we consider another direction that investigate item-item associations to analyse the dependency between items. The traditional approaches that model this association type is mainly based on similarity. Given a user, the recommendation task is to find similar items to those he adopted in the past. Significantly important, there are other dependency types captured in the item-item associations such as sequential- and correlative dependencies. The former type inferred from “sequential associations” within sequences, indicates the influence of the preceding adoptions on the following adoptions. The latter is associated with the notion of basket, which is a set of items consumed at the same time step. Associations among items in the same basket, named “basket-oriented associations”, usually imply their correlative dependencies.

In this talk, we present research works in modeling “sequential & basket-oriented associations” independently and jointly for new problem formulations of the Top-K recommendation task.

Speaker Biography

LE Duc Trong is a PhD candidate at Singapore Management University (SMU) under the supervision of Associate Professor Hady W. Lauw. He received his bachelor degree in Information Technology from University of Engineering and Technology, Vietnam National University, Hanoi. At SMU, his research topic is recommender systems, where exploits item-item associations, e.g., sequential & basket-oriented, to generate Top-K recommendations.