Reproduce “Joint Optimization of Tree-based Index and Deep Model for Recommender Systems” 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

Han Zhu, Daqing Chang, Ziru Xu, Pengye Zhang, Xiang Li, Jie He, Han Li, Jian Xu and 1 more

Joint Optimization of Tree-based Index and Deep Model (JTM) unifies the learning of tree index structures and deep user preference models for large-scale recommender systems, addressing prior objective divergence. This approach led to substantial improvements in recommendation accuracy, achieving an 11.3% increase in Click-Through Rate and a 12.9% increase in Revenue Per Mille in online A/B tests on the Taobao App.

Key insights

  • Jointly optimizing the tree index and the deep preference model under a unified objective significantly enhances recommendation accuracy by aligning their interdependent learning goals.
  • The quality of both the tree index and the user preference model is critical for effective tree-based recommendation, as an unoptimized tree can hinder performance by generating confused training samples or losing targets during search.
  • Leveraging the tree hierarchy to create level-specific user preference representations improves the predictability of user preferences and contributes directly to accuracy gains by providing contextually relevant information at each stage of retrieval.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.