Enriching user shopping history improves recommendation accuracy in e-commerce platforms.
Utilizing a multi-modal approach with pretrained image and text encoders, a robust product matching system in the fashion e-commerce industry achieves state-of-the-art results.
The author proposes a Multimodal In-Context Tuning approach, ModICT, to enhance the accuracy and diversity of product descriptions by leveraging visual and textual information.
The author proposes a Universal User Behavior Model (UBM) pre-trained on session data, utilizing contrastive learning to capture complex relations and dependencies, leading to significant performance gains across various downstream tasks.
BEQUE is a comprehensive framework designed to bridge the semantic gap in long-tail queries, enhancing e-commerce search results.