Large-scale retrieval systems must operate under strict latency constraints while maintaining high recall. Sparse retrieval offers efficiency and interpretability, whereas dense retrieval provides stronger semantic matching. Although hybrid approaches combine both signals, their interaction is often limited, especially under intersection-based retrieval. We introduce HybridSparse, an end-to-end hybrid retrieval framework that strengthens sparse--dense interaction across modeling, training, and serving. It adopts a unified encoder with a shared backbone and jointly optimizes lexical and semantic representations through co-training. To further improve alignment, we incorporate hybrid score regularization and consistency distillation, enabling more stable and effective hybrid scoring. Experiments on public benchmarks demonstrate consistent improvements over strong sparse, dense, and hybrid baselines. In large-scale production deployment for Bing advertisement retrieval, HybridSparse delivers a +1.30% RPM gain, highlighting its practical impact.

When testing data and training data come from different distributions, deep neural networks (DNNs) will face significant safety risks in practical applications. Therefore, out-of-distribution (OOD) detection techniques, which can identify OOD samples at test time and alert the system, are urgently needed. Existing graph OOD detection methods usually characterize fine-grained in-distribution (ID) patterns from multiple perspectives, and train end-to-end graph neural networks (GNNs) for prediction. However, due to the unavailability of OOD data during training, the absence of explicit supervision signals could lead to sub-optimal performance of end-to-end encoders. To address this issue, we follow thepre-training+promptingparadigm to utilize pre-trained GNN encoders, and propose Disentangled Graph Prompting (DGP), to capture fine-grained ID patterns with the help of ID graph labels. Specifically, we design two prompt generators that respectively generate class-specific and class-agnostic prompt graphs by modifying the edge weights of an input graph. We also design several effective losses to train the prompt generators and prevent trivial solutions. We conduct extensive experiments on ten datasets to demonstrate the superiority of our proposed DGP, which achieves a relative AUC improvement of 3.63% over the best graph OOD detection baseline. Ablation studies and hyper-parameter experiments further show the effectiveness of DGP.

Verbatim feedback constitutes a valuable repository of user experiences, opinions, and requirements, crucial for data engineering and software development. Extracting meaningful insights from large-scale feedback data presents a significant challenge. This paper introduces Allhands, an innovative ana-lytic framework that transforms traditional large-scale feedback analysis tasks through a natural language interface, leveraging large language models (LLMs). Allhands performs initial classification and topic modeling on feedback to convert it into a structurally augmented format, enhancing accuracy, robustness and generalization with the aid of LLMs. Subsequently, an LLM-based code-first agent interprets users' diverse natural language questions about the feedback, automatically translates them into executable call of analytic tools or code, and delivers comprehensive multi-modal responses, including text, code, tables, and images. This eliminates the need for developing individual feedback analytic tools for each request, reducing human effort and making the system more accessible and flexible to users. We evaluate Allhands across three diverse feedback datasets, demonstrating its superior efficacy in all stages of analysis, from classification and topic modeling to providing an “ask me anything” experience with comprehensive, accurate, and human-readable responses. To the best of our knowl-edge, Allhands is the first comprehensive feedback analysis framework supporting diverse and customized insight extraction requirements through a natural language interface.

User profiling is pivotal for recommendation systems, as it transforms raw user interaction data into concise and structured representations that drive personalized recommendations. While traditional embedding-based profiles lack interpretability and adaptability, recent advances with large language models (LLMs) enable text-based profiles that are semantically richer and more transparent. However, existing methods often adhere to fixed formats that limit their ability to capture the full diversity of user behaviors. In this paper, we introduce LettinGo, a novel framework for generating diverse and adaptive user profiles. By leveraging the expressive power of LLMs and incorporating direct feedback from downstream recommendation tasks, our approach avoids the rigid constraints imposed by supervised fine-tuning (SFT). Instead, we employ Direct Preference Optimization (DPO) to align the profile generator with task-specific performance, ensuring that the profiles remain adaptive and effective. LettinGo operates in three stages: (1) exploring diverse user profiles via multiple LLMs(2) evaluating profile quality based on their impact in recommendation systems, and (3) aligning the profile generation through pairwise preference data derived from task performance. Experimental results demonstrate that our framework significantly enhances recommendation accuracy, flexibility, and contextual awareness. This work enhances profile generation as a key innovation for next-generation recommendation systems.

Owing to the unprecedented capability in semantic understanding and logical reasoning, large language models (LLMs) have shown fantastic potential in developing next-generation sequential recommender systems (RSs). However, existing LLM-based sequential RSs mostly separate index generation from sequential recommendation, leading to insufficient integration between semantic information and collaborative information. On the other hand, the neglect of user-related information hinders LLM-based sequential RSs from exploiting high-order user-item interaction patterns. In this paper, we propose the End-to-End Dual Dynamic (ED2) recommender, the first LLM-based sequential RS which adopts dual dynamic index mechanism, targeting resolving the above limitations simultaneously. The dual dynamic index mechanism can not only assembly index generation and sequential recommendation into a unified LLM-backbone pipeline, but also make it practical for LLM-based sequential recommender to take advantage of user-related information. Specifically, to facilitate the LLM comprehension ability to dual dynamic index, we propose a multigrained token regulator which constructs alignment supervision based on LLMs semantic knowledge across multiple representation granularities. Moreover, the associated user collection data and a series of novel instruction tuning tasks are specially customized to capture the high-order user-item interaction patterns. Extensive experiments on three public datasets demonstrate the superiority of ED2, achieving an average improvement of 19.62% in Hit-Rate and 21.11% in NDCG.

We address the real problem of safe, robust, adaptive resource oversubscription in uncertain environments with our proposed novel technique of chance-constrained imitation learning. Our objective is to enhance resource efficiency while ensuring safety against congestion risk. Traditional supervised or forecasting models are ineffective in learning adaptive oversubscription policies, and conventional online optimization or reinforcement learning is difficult to deploy on real systems. Offline policy learning methods, such as Imitation Learning (IL) can leverage historical resource utilization telemetry data to learn effective policies if we can ensure robustness and safety from the underlying uncertainty in the domain, and thus the data. Our work investigates the nature of this uncertainty, how it can be quantified and proposes a novel chance-constrained IL that implicitly models such uncertainty in a principled manner via additional knowledge in the form of stochastic constraints on the associated risk, to learn provably safe and robust policies. We show empirically a substantial improvement (~ 3-4×) in capacity efficiency and congestion safety in test as well as real deployments.

As the non-fungible token (NFT) market flourishes, price prediction emerges as a pivotal direction for investors gaining valuable insight to maximize returns. However, existing works suffer from a lack of practical definitions and standardized evaluations, limiting their practical application. Moreover, the influence of users' multi-behaviour transactions that are publicly accessible on NFT price is still not explored and exhibits challenges. In this paper, we address these gaps by presenting a practical and hierarchical problem definition. This approach unifies both collection-level and token-level task and evaluation methods, which cater to varied practical requirements of investors. To further understand the impact of user behaviours on the variation of NFT price, we propose a general wallet profiling framework and develop a COmmunity enhanced Multi-bEhavior Transaction graph model, named COMET. COMET profiles wallets with a comprehensive view and considers the impact of diverse relations and interactions within the NFT ecosystem on NFT price variations, thereby improving prediction performance. Extensive experiments conducted in our deployed system demonstrate the superiority of COMET, underscoring its potential in the insight toolkit for NFT investors.

Deploying pre-trained transformer models like BERT on downstream tasks in resource-constrained scenarios is challenging due to their high inference cost, which grows rapidly with input sequence length. In this work, we propose a constraint-aware and ranking-distilled token pruning method ToP, which selectively removes unnecessary tokens as input sequence passes through layers, allowing the model to improve online inference speed while preserving accuracy. ToP overcomes the limitation of inaccurate token importance ranking in the conventional self-attention mechanism through a ranking-distilled token distillation technique, which distills effective token rankings from the final layer of unpruned models to early layers of pruned models. Then, ToP introduces a coarse-to-fine pruning approach that automatically selects the optimal subset of transformer layers and optimizes token pruning decisions within these layers through improved L0 regularization. Extensive experiments on GLUE benchmark and SQuAD tasks demonstrate that ToP outperforms state-of-the-art token pruning and model compression methods with improved accuracy and speedups. ToP reduces the average FLOPs of BERT by 8.1X while achieving competitive accuracy on GLUE, and provides a real latency speedup of up to 7.4X on an Intel CPU. Code is available at https://github.com/microsoft/Moonlit/tree/main/ToP

The nucleus of online sponsored search systems lies in measuring the relevance between the search intents of users and the advertising purposes of advertisers. Existing conventional doublet-based (query-keyword) relevance models solely rely on short queries and keywords to uncover such intents, which ignore the diverse and personalized preferences of participants (i.e., users and advertisers), resulting in undesirable advertising performance. In this paper, we investigate the novel problem of Personalized A dvertiser-aware Sponsored Search (PASS). Our motivation lies in incorporating the portraits of users and advertisers into relevance models to facilitate the modeling of intrinsic search intents and advertising purposes, leading to a quadruple-based (i.e., user-query-keyword-advertiser) task. Various types of historical behaviors are explored in the format of hypergraphs to provide abundant signals on identifying the preferences of participants. A novel heterogeneous textual hypergraph transformer is further proposed to deeply fuse the textual semantics and the high-order hypergraph topology. Our proposal is extensively evaluated over real industry datasets, and experimental results demonstrate its superiority.

Although deep pre-trained language models have shown promising benefit in a large set of industrial scenarios, including Click-Through-Rate (CTR) prediction, how to integrate pre-trained language models that handle only textual signals into a prediction pipeline with non-textual features is challenging.

Relevance models measure the semantic closeness between queries and the candidate ads, widely recognized as the nucleus of sponsored search systems. Conventional relevance models solely rely on the textual data within the queries and ads, whose performance is hindered by the scarce semantic information in these short texts. Recently, user behavior graphs have been incorporated to provide complementary information beyond pure textual semantics.Despite the promising performance, behavior-enhanced models suffer from exhausting resource costs due to the extra computations introduced by explicit topological aggregations. In this paper, we propose a novel Multi-Grained Topological Pre-Training paradigm, MGTLM, to teach language models to understand multi-grained topological information in behavior graphs, which contributes to eliminating explicit graph aggregations and avoiding information loss. Extensive experimental results over online and offline settings demonstrate the superiority of our proposal.

Multi-lingual text advertisement generation is a critical task for international companies, such as Microsoft. Due to the lack of training data, scaling out text advertisements generation to low-resource languages is a grand challenge in the real industry setting. Although some methods transfer knowledge from rich-resource languages to low-resource languages through a pre-trained multi-lingual language model, they fail in balancing the transferability from the source language and the smooth expression in target languages. In this paper, we propose a unified Self-Supervised Augmentation and Generation (SAG) architecture to handle the multi-lingual text advertisements generation task in a real production scenario. To alleviate the problem of data scarcity, we employ multiple data augmentation strategies to synthesize training data in target languages. Moreover, a self-supervised adaptive filtering structure is developed to alleviate the impact of the noise in the augmented data. The new state-of-the-art results on a well-known benchmark verify the effectiveness and generalizability of our proposed framework, and deployment in Microsoft Bing demonstrates the superior performance of our method.

As the fundamental basis of sponsored search, relevance modeling measures the closeness between the input queries and the candidate ads. Conventional relevance models solely rely on the textual data, which suffer from the scarce semantic signals within the short queries. Recently, user historical click behaviors are incorporated in the format of click graphs to provide additional correlations beyond pure textual semantics, which contributes to advancing the relevance modeling performance. However, user behaviors are usually arbitrary and unpredictable, leading to the noisy and sparse graph topology. In addition, there exist other types of user behaviors besides clicks, which may also provide complementary information. In this paper, we study the novel problem of heterogeneous behavior graph learning to facilitate relevance modeling task. Our motivation lies in learning an optimal and task-relevant heterogeneous behavior graph consisting of multiple types of user behaviors. We further propose a novel HBGLR model to learn the behavior graph structure by mining the sophisticated correlations between node semantics and graph topology, and encode the textual semantics and structural heterogeneity into the learned representations. Our proposal is evaluated over real-world industry datasets, and has been mainstreamed in the Bing ads. Both offline and online experimental results demonstrate its superiority.

In sponsored search engines, pre-trained language models have shown promising performance improvements on Click-Through-Rate (CTR) prediction. A widely used approach for utilizing pre-trained language models in CTR prediction consists of fine-tuning the language models with click labels and early stopping on peak value of the obtained Area Under the ROC Curve (AUC). Thereafter the output of these fine-tuned models, i.e., the final score or intermediate embedding generated by language model, is used as a new Natural Language Processing (NLP) feature into CTR prediction baseline. This cascade approach avoids complicating the CTR prediction baseline, while keeping flexibility and agility. However, we show in this work that calibrating separately the language model based on the peak single model AUC does not always yield NLP features that give the best performance in CTR prediction model ultimately. Our analysis reveals that the misalignment is due to overlap and redundancy between the new NLP features and the existing features in CTR prediction baseline. In other words, the NLP features can improve CTR prediction better if such overlap can be reduced.

Embedding based retrieval (EBR) is a fundamental building block in many web applications. However, EBR in sponsored search is distinguished from other generic scenarios and technically challenging due to the need of serving multiple retrieval purposes: firstly, it has to retrieve high-relevance ads, which may exactly serve user's search intent; secondly, it needs to retrieve high-CTR ads so as to maximize the overall user clicks. In this paper, we present a novel representation learning framework Uni-Retriever developed for Bing Search, which unifies two different training modes knowledge distillation and contrastive learning to realize both required objectives. On one hand, the capability of making high-relevance retrieval is established by distilling knowledge from the "relevance teacher model''. On the other hand, the capability of making high-CTR retrieval is optimized by learning to discriminate user's clicked ads from the entire corpus. The two training modes are jointly performed as a multi-objective learning process, such that the ads of high relevance and CTR can be favored by the generated embeddings. Besides the learning strategy, we also elaborate our solution for EBR serving pipeline built upon the substantially optimized DiskANN, where massive-scale EBR can be performed with competitive time and memory efficiency, and accomplished in high-quality. We make comprehensive offline and online experiments to evaluate the proposed techniques, whose findings may provide useful insights for the future development of EBR systems. Uni-Retriever has been mainstreamed as the major retrieval path in Bing's production thanks to the notable improvements on the representation and EBR serving quality.

Learning informative representations of users and items from the historical interactions is crucial to collaborative filtering (CF). Existing CF approaches usually model interactions solely within the Euclidean space. However, the sophisticated user-item interactions inherently present highly non-Euclidean anatomy with various types of geometric patterns (i.e., tree-likeness and cyclic structures). The Euclidean-based models may be inadequate to fully uncover the intent factors beneath such hybrid-geometry interactions. To remedy this deficiency, in this paper, we study the novel problem of Geometric Disentangled Collaborative Filtering (GDCF), which aims to reveal and disentangle the latent intent factors across multiple geometric spaces. A novel generative GDCF model is proposed to learn geometric disentangled representations by inferring the high-level concepts associated with user intentions and various geometries. Empirically, our proposal is extensively evaluated over five real-world datasets, and the experimental results demonstrate the superiority of GDCF.

Ad-hoc search calls for the selection of appropriate answers from a massive-scale corpus. Nowadays, the embedding-based retrieval (EBR) becomes a promising solution, where deep learning based document representation and ANN search techniques are allied to handle this task. However, a major challenge is that the ANN index can be too large to fit into memory, given the considerable size of answer corpus. In this work, we tackle this problem with Bi-Granular Document Representation, where the lightweight sparse embeddings are indexed and standby in memory for coarse-grained candidate search, and the heavyweight dense embeddings are hosted in disk for fine-grained post verification. For the best of retrieval accuracy, a Progressive Optimization framework is designed. The sparse embeddings are learned ahead for high-quality search of candidates. Conditioned on the candidate distribution induced by the sparse embeddings, the dense embeddings are continuously learned to optimize the discrimination of ground-truth from the shortlisted candidates. Besides, two techniques: the contrastive quantization and the locality-centric sampling are introduced for the learning of sparse and dense embeddings, which substantially contribute to their performances. Thanks to the above features, our method effectively handles massive-scale EBR with strong advantages in accuracy: with up to recall gain on million-scale corpus, and up to recall gain on billion-scale corpus. Besides, Our method is applied to a major sponsored search platform with substantial gains on revenue (), Recall () and CTR (). Our code is available at https://github.com/microsoft/BiDR.

Sponsored search ads appear next to search results when people look for products and services on search engines. In recent years, they have become one of the most lucrative channels for marketing. As the fundamental basis of search ads, relevance modeling has attracted increasing attention due to the significant research challenges and tremendous practical value. Most existing approaches solely rely on the semantic information in the input query-ad pair, while the pure semantic information in the short ads data is not sufficient to fully identify user's search intents. Our motivation lies in incorporating the tremendous amount of unsupervised user behavior data from the historical search logs as the complementary graph to facilitate relevance modeling. In this paper, we extensively investigate how to naturally fuse the semantic textual information with the user behavior graph, and further propose three novel AdsGNN models to aggregate topological neighborhood from the perspectives of nodes, edges and tokens. Furthermore, two critical but rarely investigated problems, domain-specific pre-training and long-tail ads matching, are studied thoroughly. Empirically, we evaluate the AdsGNN models over the large industry dataset, and the experimental results of online/offline tests consistently demonstrate the superiority of our proposal.

Anomaly detection on multivariate time-series is of great importance in both data mining research and industrial applications. Recent approaches have achieved significant progress in this topic, but there is remaining limitations. One major limitation is that they do not capture the relationships between different time-series explicitly, resulting in inevitable false alarms. In this paper, we propose a novel self-supervised framework for multivariate time-series anomaly detection to address this issue. Our framework considers each univariate time-series as an individual feature and includes two graph attention layers in parallel to learn the complex dependencies of multivariate time-series in both temporal and feature dimensions. In addition, our approach jointly optimizes a forecasting-based model and a reconstruction-based model, obtaining better time-series representations through a combination of single-timestamp prediction and reconstruction of the entire time-series. We demonstrate the efficacy of our model through extensive experiments. The proposed method outperforms other state-of-the-art models on three real-world datasets. Further analysis shows that our method has good interpretability and is useful for anomaly diagnosis.

Large companies need to monitor various metrics (for example, Page Views and Revenue) of their applications and services in real time. At Microsoft, we develop a time-series anomaly detection service which helps customers to monitor the time-series continuously and alert for potential incidents on time. In this paper, we introduce the pipeline and algorithm of our anomaly detection service, which is designed to be accurate, efficient and general. The pipeline consists of three major modules, including data ingestion, experimentation platform and online compute. To tackle the problem of time-series anomaly detection, we propose a novel algorithm based on Spectral Residual (SR) and Convolutional Neural Network (CNN). Our work is the first attempt to borrow the SR model from visual saliency detection domain to time-series anomaly detection. Moreover, we innovatively combine SR and CNN together to improve the performance of SR model. Our approach achieves superior experimental results compared with state-of-the-art baselines on both public datasets and Microsoft production data.

We propose a novel training scheme for fast matching models in Search Ads, motivated by practical challenges. The first challenge stems from the pursuit of high throughput, which prohibits the deployment of inseparable architectures, and hence greatly limits model accuracy. The second problem arises from the heavy dependency on human provided labels, which are expensive and time-consuming to collect, yet how to leverage unlabeled search log data is rarely studied. The proposed training framework targets on mitigating both issues, by treating the stronger but undeployable models as annotators, and learning a deployable model from both human provided relevance labels and weakly annotated search log data. Specifically, we first construct multiple auxiliary tasks from the enumerated relevance labels, and train the annotators by jointly learning from those related tasks. The annotation models are then used to assign scores to both labeled and unlabeled training samples. The deployable model is firstly learnt on the scored unlabeled data, and then fine-tuned on scored labeled data, by leveraging both labels and scores via minimizing the proposed label-aware weighted loss. According to our experiments, compared with the baseline that directly learns from relevance labels, training by the proposed framework outperforms it by a large margin, and improves data efficiency substantially by dispensing with 80% labeled samples. The proposed framework allows us to improve the fast matching model by learning from stronger annotators while keeping its architecture unchanged. Meanwhile, it offers a principled manner to leverage search log data in the training phase, which could effectively alleviate our dependency on human provided labels.

Ad click prediction is a task to estimate the click-through rate (CTR) in sponsored ads, the accuracy of which impacts user search experience and businesses' revenue. State-of-the-art sponsored search systems typically model it as a classification problem and employ machine learning approaches to predict the CTR per ad. In this paper, we propose a new approach to predict ad CTR in sequence which considers user browsing behavior and the impact of top ads quality to the current one. To the best of our knowledge, this is the first attempt in the literature to predict ad CTR by using Recurrent Neural Networks (RNN) with Long Short-Term Memory (LSTM) cells. The proposed model is evaluated on a real dataset and we show that LSTM-RNN outperforms DNN model on both AUC and RIG. Since the RNN inference is time consuming, a simplified version is also proposed, which can achieve more than half of the gain with the overall serving cost almost unchanged.