A Stacking and Transfer Learning with Diverse Similarity For Building Multilingual Session-based Recommendation Systems

Jiangwei Luo, Zhouzhou He, Ye Tang, Wentao Tang, Li Cheng

Keywords: Session-based Recommendation, Transfer Learning, KDD Cup, Multilingual Recommendation Systems

Abstract:

It is crucial for e-commerce stores to simulate customers’ shopping intentions, as it directly affects user experience and user engagement. Session-based recommendations, which utilizes customer session data to predict their next purchase, has become increasingly popular with advances in data mining and machine learning techniques.However, few studies have explored session-based recommendation under real-world multilingual and imbalanced scenarios. To imporve this research, Amazon published a large-scale shopping queries dataset and hosted KDD Cup 2023 Challenge for building multilingual recommendation systems.
In this pager, the recommendation algorithm team of MGTV present an effective and industrial solution to this challenge,our recommendation pipeline is composed of four stages, which is focused on data preprocessing, candidate generation, build ranking models and blend. Finally, with our solution, our team MGTV_REC won 2nd place in task1 and task2 among 2008 participants.

Submission Number: 12