Xinyu Qian

Research Areas

Authored Publications
Sort By
  • Title
  • Title, descending
  • Year
  • Year, descending
    Preview abstract We describe how we built three recommendation products from scratch at Google Chrome Web Store, namely context-based recommendations, related extension recommendations, and personalized recommendations. Unlike most existing papers that focus on novel algorithms, this paper focuses on sharing practical experiences building large scale recommender systems under various real-world constraints, such as privacy constraints, data sparsity issues, highly skewed data distribution, and product design choices, such as user interface. We show how these constraints make standard approaches difficult to succeed in practice. We share success stories that turn very negative live metrics to very positive, by introducing 1) how we use interpretable neural models to bootstrap the systems, helps identifying pipeline issues, and paves way for more advanced models. 2) A new item-item based recommendation algorithm that works under highly skewed data distributions, and 3) how two products can help bootstrapping the third one, which significantly reduces development cycles and bypasses various real-world difficulties. All the explorations in this work are verified in live traffic on millions of users. We believe the findings in this work can help practitioners to bootstrap and build large-scale recommender systems. View details
    Preview abstract A well-known challenge in leveraging implicit user feedback like clicks to improve real-world search services and recommender systems is its inherent bias. Most existing click models are based on the examination hypothesis in user behaviors and differ in how to model such an examination bias. However, they are constrained by assuming a simple position-based bias or enforcing a sequential order in user examination behaviors. These assumptions are insufficient to capture complex real-world user behaviors and hardly generalize to modern user interfaces (UI) in web applications (e.g., results shown in a grid view). In this work, we propose a fully data-driven neural model for the examination bias, Cross-Positional Attention (XPA), which is more flexible in fitting complex user behaviors. Our model leverages the attention mechanism to effectively capture cross-positional interactions among displayed items and is applicable to arbitrary UIs. We employ XPA in a novel neural click model that can both predict clicks and estimate relevance. Our experiments on offline synthetic data sets show that XPA is robust among different click generation processes. We further apply XPA to a large-scale real-world recommender system, showing significantly better results than baselines in online A/B experiments that involve millions of users. This validates the necessity to model more complex user behaviors than those proposed in the literature. View details