Retrieval augmentation, which enhances downstream models by a knowledge retriever and an external corpus instead of by merely increasing the number of model parameters, has been successfully applied to many natural language processing(NLP) tasks such as text classification, question answering and so on. However, existing methods that separately or asynchronously train the retriever and downstream model mainly due to the non-differentiability between the two parts, usually lead to degraded performance compared to end-to-end joint training. In this paper, we propose Differentiable Retrieval Augmentation via Generative lANguage modeling(Dragan), to address this problem by a novel differentiable reformulation. We demonstrate the effectiveness of our proposed method on a challenging NLP task in e-commerce search, namely query intent classification. Both the experimental results and ablation study show that the proposed method significantly and reasonably improves the state-of-the-art baselines on both offline evaluation and online A/B test.

Ranking model plays an essential role in e-commerce search and recommendation. An effective ranking model should give a personalized ranking list for each user according to the user preference. Existing algorithms usually extract a user representation vector from the user behavior sequence, then feed the vector into a feed-forward network (FFN) together with other features for feature interactions, and finally produce a personalized ranking score. Despite tremendous progress in the past, there is still room for improvement. Firstly, the personalized patterns of feature interactions for different users are not explicitly modeled. Secondly, most of existing algorithms have poor personalized ranking results for long-tail users with few historical behaviors due to the data sparsity.To overcome the two challenges, we propose Attention Weighted Mixture of Experts (AW-MoE) with contrastive learning for personalized ranking. Firstly, AW-MoE leverages the MoE framework to capture personalized feature interactions for different users. To model the user preference, the user behavior sequence is simultaneously fed into expert networks and the gate network. Within the gate network, one gate unit and one activation unit are designed to adaptively learn the fine-grained activation vector for experts using an attention mechanism. Secondly, a random masking strategy is applied to the user behavior sequence to simulate long-tail users, and an auxiliary contrastive loss is imposed to the output of the gate network to improve the model generalization for these users. This is validated by a higher performance gain on the long-tail user test set.Experiment results on a JD real production dataset and a public dataset demonstrate the effectiveness of AW-MoE, which significantly outperforms state-of-art methods. Notably, AW-MoE has been successfully deployed in the JD e-commerce search engine, serving the real traffic of hundreds of millions of active users.

Recent years have seen a significant amount of interests in Sequential Recommendation (SR), which aims to understand and model the sequential user behaviors and the interactions between users and items over time. Surprisingly, despite the huge success Sequential Recommendation has achieved, there is little study on Sequential Search (SS), a twin learning task that takes into account a user's current and past search queries, in addition to behavior on historical query sessions. The SS learning task is even more important than the counterpart SR task for most of E-commence companies due to its much larger online serving demands as well as traffic volume.

Product search is the most common way for people to satisfy their shopping needs on e-commerce websites. Products are typically annotated with one of several broad categorical tags, such as "Clothing" or "Electronics", as well as finer-grained categories like "Refrigerator" or "TV", both under "Electronics". These tags are used to construct a hierarchy of query categories. Distributions of features such as price and brand popularity vary wildly across query categories. In addition, feature importance for the purpose of CTR/CVR predictions differs from one category to another. In this work, we leverage the Mixture of Expert (MoE) framework to learn a ranking model that specializes for each query category. In particular, our gate network relies solely on the category ids extracted from the user query.While classical MoE's pick expert towers spontaneously for each input example, we explore two techniques to establish more explicit and transparent connections between the experts and query categories. To help differentiate experts on their domain specialties, we introduce a form of adversarial regularization among the expert outputs, forcing them to disagree with one another. As a result, they tend to approach each prediction problem from different angles, rather than copying one another. This is validated by a much stronger clustering effect of the gate output vectors under different categories. In addition, soft gating constraints based on the categorical hierarchy are imposed to help similar products choose similar gate values. and make them more likely to share similar experts. This allows aggregation of training data among smaller sibling categories to overcome data scarcity.Experiments on a learning-to-rank dataset collected from the JD e-commerce search log demonstrate that MoE with these improvements consistently outperforms competing models, in terms of offline metrics and online AB tests.

We propose a novel domain-specific generative pre-training (DSGPT) method for text generation and apply it to the product title and review summarization problems on E-commerce mobile display. First, we adopt a decoder-only transformer architecture, which fits well for fine-tuning tasks by combining input and output all together. Second, we demonstrate utilizing only small amount of pre-training data in related domains is powerful. Pre-training a language model from a general corpus such as Wikipedia or the Common Crawl requires tremendous time and resource commitment, and can be wasteful if the downstream tasks are limited in variety. Our DSGPT is pre-trained on a limited dataset, the Chinese short text summarization dataset (LCSTS). Third, our model does not require product-related human-labeled data. For title summarization task, the state of art explicitly uses additional background knowledge in training and predicting stages. In contrast, our model implicitly captures this knowledge and achieves significant improvement over other methods, after fine-tuning on the public Taobao.com dataset. For review summarization task, we utilize JD.com in-house dataset, and observe similar improvement over standard machine translation methods which lack the flexibility of fine-tuning. Our proposed work can be simply extended to other domains for a wide range of text generation tasks.

Embedding index that enables fast approximate nearest neighbor(ANN) search, serves as an indispensable component for state-of-the-art deep retrieval systems. Traditional approaches, often separating the two steps of embedding learning and index building, incur additional indexing time and decayed retrieval accuracy. In this paper, we propose a novel method called Poeem, which stands for product quantization based embedding index jointly trained with deep retrieval model, to unify the two separate steps within an end-to-end training, by utilizing a few techniques including the gradient straight-through estimator, warm start strategy, optimal space decomposition and Givens rotation. Extensive experimental results show that the proposed method not only improves retrieval accuracy significantly but also reduces the indexing time to almost none. We have open sourced our approach for the sake of comparison and reproducibility.

Relevance has significant impact on user experience and business profit for e-commerce search platform. In this work, we propose a data-driven framework for search relevance prediction, by distilling knowledge from BERT and related multi-layer Transformer teacher models into simple feed-forward networks with large amount of unlabeled data. The distillation process produces a student model that recovers more than 97% test accuracy of teacher models on new queries, at a serving cost that's several magnitude lower (latency 150x lower than BERT-Base and 15x lower than the most efficient BERT variant, TinyBERT). The applications of temperature rescaling and teacher model stacking further boost model accuracy, without increasing the student model complexity. We present experimental results on both in-house e-commerce search relevance data as well as a public data set on sentiment analysis from the GLUE benchmark. The latter takes advantage of another related public data set of much larger scale, while disregarding its potentially noisy labels. Embedding analysis and case study on the in-house data further highlight the strength of the resulting model. By making the data processing and model training source code public, we hope the techniques presented here can help reduce energy consumption of the state of the art Transformer models and also level the playing field for small organizations lacking access to cutting edge machine learning hardwares.

Nowadays e-commerce search has become an integral part of many people's shopping routines. Two critical challenges stay in today's e-commerce search: how to retrieve items that are semantically relevant but not exact matching to query terms, and how to retrieve items that are more personalized to different users for the same search query. In this paper, we present a novel approach called DPSR, which stands for Deep Personalized and Semantic Retrieval, to tackle this problem. Explicitly, we share our design decisions on how to architect a retrieval system so as to serve industry-scale traffic efficiently and how to train a model so as to learn query and item semantics accurately. Based on offline evaluations and online A/B test with live traffics, we show that DPSR model outperforms existing models, and DPSR system can retrieve more personalized and semantically relevant items to significantly improve users' search experience by +1.29% conversion rate, especially for long tail queries by +10.03%. As a result, our DPSR system has been successfully deployed into JD.com's search production since 2019.

We introduce deep learning models to the two most important stages in product search at JD.com, one of the largest e-commerce platforms in the world. Specifically, we outline the design of a deep learning system that retrieves semantically relevant items to a query within milliseconds, and a pairwise deep re-ranking system, which learns subtle user preferences. Compared to traditional search systems, the proposed approaches are better at semantic retrieval and personalized ranking, achieving significant improvements.

We introduce simple, efficient algorithms for computing a MinHash of a probability distribution, suitable for both sparse and dense data, with equivalent running times to the state of the art for both cases. The collision probability of these algorithms is a new measure of the similarity of positive vectors which we investigate in detail. We describe the sense in which this collision probability is optimal for any Locality Sensitive Hash based on sampling. We argue that this similarity measure is more useful for probability distributions than the similarity pursued by other algorithms for weighted MinHash, and is the natural generalization of the Jaccard index.