Reproduce “Towards Sustainable Growth: A Multi-Value-Aware Retrieval Framework for E-Commerce Search” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yifan Wang, Yixuan Wang, YiDan Liang, Qiang Liu, Fei Xiao
The research from Alibaba's Taobao & Tmall Group introduces GrowthGR, a retrieval framework that uses causal inference and a generative model to balance immediate conversion with the long-term growth potential of new items in e-commerce search. It demonstrated a 5.39% increase in new item GMV and a 0.31% gain in overall Search GMV during online A/B tests on Taobao.
Key insights
- Quantifying long-term item growth potential can be achieved by modeling the causal impact of individual user interactions on an item's future transaction trajectory.
- Generative retrieval, utilizing hierarchical semantic IDs and Transformer architectures, can effectively address data sparsity for new items and generalize semantic relationships for broader discovery.
- A multi-objective optimization strategy that explicitly balances immediate conversion with predicted long-term growth, while mitigating popularity bias, is crucial for fostering sustainable e-commerce ecosystem health.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.