Large language models (LLMs) have shown great potential in enhancing search and recommender systems by providing rich semantic representations from unstructured texts. However, directly integrating LLM embeddings into industrial recommendation pipelines often results in subpar performance due to the semantic and distributional mismatch between pre-trained LLM features and domain-specific, feedback-driven representations. Existing approaches struggle to effectively align LLM embeddings with recommendation objectives, often facing challenges such as label misalignment or the potential loss of semantic diversity during fine-tuning. In this work, we present TreeBridge, a novel framework that introduces a structure-aware generative encoding tree to bridge the semantic gap between LLM embeddings and recommendation tasks. It preserves the external semantic richness of LLM embeddings, while learning label-informed structures that capture user preferences and interaction patterns. This enables the generation of task-adaptive representations without compromising embedding diversity. We further adopt an online-offline hybrid service paradigm to ensure low-latency real-world deployment. TreeBridge has been deployed on the Shopee e-commerce platform, one of the largest online shopping platforms in Southeast Asia serving hundreds of millions of users. Since its deployment in May 2025, it has helped the company achieve a commercially significant 1.55% relative improvement in gross merchandise volume (GMV). The deployment experience demonstrates the effectiveness, scalability, and significant commercial value of TreeBridge.

Embedding-based collaborative filtering, often coupled with nearest neighbor search, is widely deployed in large-scale recommender systems for personalized content selection. Modern systems leverage multiple implicit feedback signals (e.g., clicks, add to cart, purchases) to model user preferences comprehensively. However, prevailing approaches adopt a feedback-wise modeling paradigm, which (1) fails to capture the structured progression of user engagement entailed among different feedback and (2) embeds feedback-specific information into disjoint spaces, making representations incommensurable, increasing system complexity, and leading to suboptimal retrieval performance. A promising alternative is Ordinal Logistic Regression (OLR), which explicitly models discrete ordered relations. However, existing OLR-based recommendation models mainly focus on explicit feedback (e.g., movie ratings) and struggle with implicit, correlated feedback, where ordering is vague and non-linear. Moreover, standard OLR lacks flexibility in handling feedback-dependent covariates, resulting in suboptimal performance in real-world systems. To address these limitations, we propose Generalized Neural Ordinal Logistic Regression (GNOLR), which encodes multiple feature-feedback dependencies into a unified, structured embedding space and enforces feedback-specific dependency learning through a nested optimization framework. Thus, GNOLR enhances predictive accuracy, captures the progression of user engagement, and simplifies the retrieval process. We establish a theoretical comparison with existing paradigms, demonstrating how GNOLR avoids disjoint spaces while maintaining effectiveness. Extensive experiments on ten real-world datasets show that GNOLR significantly outperforms state-of-the-art methods in efficiency and adaptability.

Maximum Inner Product Search (MIPS) is a fundamental challenge in machine learning and information retrieval, particularly in high-dimensional data applications. Existing approaches to MIPS either rely solely on Inner Product (IP) similarity, which faces issues with local optima and redundant computations, or reduce the MIPS problem to the Nearest Neighbor Search under the Euclidean metric via space projection, leading to topology destruction and information loss. Despite the divergence of the two paradigms, we argue that there is no inherent binary opposition between IP and Euclidean metrics. By stitching IP and Euclidean in the design of indexing and search algorithms, we can significantly enhance MIPS performance. Specifically, this paper explores the theoretical and empirical connections between these two metrics from the MIPS perspective. Our investigation, grounded in graph-based search, reveals that different indexing and search strategies offer distinct advantages for MIPS, depending on the underlying data topology. Building on these insights, we introduce a novel graph-based index called Metric-Amphibious Graph (MAG) and a corresponding search algorithm, Adaptive Navigation with Metric Switch (ANMS). To facilitate parameter tuning for optimal performance, we identify three statistical indicators that capture essential data topology properties and correlate strongly with parameter tuning. Extensive experiments on 12 real-world datasets demonstrate that MAG outperforms existing state-of-the-art methods, achieving up to 4x search speedup while maintaining adaptability and scalability.

Maximum Inner Product Search (MIPS) for high-dimensional vectors is pivotal across databases, information retrieval, and artificial intelligence. Existing methods either reduce MIPS to Nearest Neighbor Search (NNS) while suffering from harmful vector space transformations, or attempt to tackle MIPS directly but struggle to mitigate redundant computations due to the absence of the triangle inequality. This paper presents a novel theoretical framework that equates MIPS with NNS without requiring space transformation, thereby allowing us to leverage advanced graph-based indices for NNS and efficient edge pruning strategies, significantly reducing unnecessary computations. Despite a strong baseline set by our theoretical analysis, we identify and address two persistent challenges to further refine our method: the introduction of the P roximity Graph with S pherical P athway (PSP), designed to mitigate the issue of MIPS solutions clustering around large-norm vectors, and the implementation of A daptive E arly T ermination (AET), which efficiently curtails the excessive exploration once an accuracy bottleneck is reached. Extensive experiments reveal that our method is superior to existing state-of-the-art techniques in search efficiency, scalability, and practical applicability. Compared with state-of-the-art graph-based methods, it achieves an average 35% speed-up in query processing and a 3× reduction in index size. Notably, our approach has been validated and deployed in the search engines of Shopee, a well-known online shopping platform. Our code and an industrial-scale dataset for offline evaluation will also be released to address the absence of e-commerce data in public benchmarks.

This paper presents a novel calibration scheme for single-channel SAR, pipelined and pipelined-SAR ADCs using Kolmogorov–Arnold networks (KANs). In the proposed scheme, a multi-sample KAN (MS-KAN) is designed to realize nonlinear code synthesis (NLCS), achieving effective calibration for general nonlinear errors. The MS-KAN-based calibrator can be converted into an analytical expression, making the calibration process transparent, with stronger interpretability, predictability and reliability compared to previous neural network-based calibration algorithms, and assisting in the analysis of ADC nonidealities. Meanwhile, the proposed scheme achieves high calibration performance with low hardware overhead. The proposed scheme also requires much fewer training samples, thereby reducing the effort required for both chip testing and network training. The MS-KAN-based calibrator is verified with two silicon-proven ADCs, a 14-bit 1.3 GS/s pipelined ADC and a 10-bit 700MS/s SAR ADC. Measurement results show that SFDR is improved by 11.5 dB to 30.9 dB after calibration. The quantized calibrators are implemented on both FPGA and 28nm CMOS technology, where a piecewise polynomial (PWP) method is adopted to simplify the implementation of the calibrator. The post-layout simulation results show that the calibrator for the real-time calibration of the pipelined ADC consumes only 6.32 mW, while the calibrator for the SAR ADC consumes 2.42 mW.

In recent years, recommender systems have advanced rapidly, where embedding learning for users and items plays a critical role. A standard method learns a unique embedding vector for each user and item. However, such a method has two important limitations in real-world applications: (1) it is hard to learn embeddings that generalize well for users and items that have rare interactions, and (2) it may incur unbearably high memory costs when the number of users and items scales up. Existing approaches either can only address one of the limitations or have flawed overall performances. In this article, we propose Clustered Embedding Learning (CEL) as an integrated solution to these two problems. CEL is a plug-and-play embedding learning framework that can be combined with any differentiable feature interaction model. It is capable of achieving improved performance, especially for cold users and items, with reduced memory cost. CEL enables automatic and dynamic clustering of users and items in a top-down fashion, where clustered entities could jointly learn a shared embedding. The accelerated version of CEL has an optimal time complexity, which supports efficient online updates. Theoretically, we prove the identifiability and the existence of a unique optimal number of clusters for CEL in the context of nonnegative matrix factorization. Empirically, we validate the effectiveness of CEL on three public datasets and one business dataset, showing its consistently superior performance against state-of-the-art methods. In particular, when incorporating CEL into the business model, it brings an improvement of \(+0.6\%\) in AUC, which translates into a significant revenue gain; meanwhile, the size of the embedding table gets 2,650 times smaller. Additionally, we demonstrate that if there is enough memory, learning a personalized embedding for each user and item around their clustering centers is feasible and can further boost performance. In this article, we enhance and extend the personalization technique we initially proposed in our earlier work [ 4 ], which introduced an offset regularization to prevent personalized embeddings from drifting too far away from the central (cluster) embedding, thereby mitigating overfitting. However, in [ 4 ], we simply applied a uniform regularization weight across all embeddings, which, given the considerable variation in the number of their associated interactions, is suboptimal. To address this, we investigate in this article the strategies for non-uniform offset regularization that adjusts regularization weights according to the number of associated interactions, which leads to significant improvements compared with uniform offset regularization. Furthermore, we extend CEL into Meta-CEL, factoring in future personalization during cluster optimization, which leads to additional enhancements in personalization performance.

This article discusses a 12-bit 2-GS/s pipeline analog-to-digital converter (ADC). A self-calibration technique is employed to correct linear errors due to capacitor mismatches and interstage gain errors (IGEs). To counteract the effects of power supply and temperature variations, the first three stages of the ADC are equipped with least-mean-squares (LMS) IGE background calibrations, enhanced by the injection of a 1-bit dither into these stages. The computational engines designed for background calibration were reused for self-calibration, simplifying the overall design. An improved integrated input buffer drives the ADC, achieving a bandwidth of approximately 6.3 GHz, which is essential for high-speed data acquisition and processing. Moreover, a low-power operational transconductance amplifier (OTA) and reference buffer, both operating on a 1.0-V supply, are implemented to minimize the chip’s power consumption. The 12-bit pipeline prototype ADC, fabricated using a 28-nm CMOS process, operates at 2-GS/s with a 1.0-Vpp input signal. It delivers a signal-to-noise-and-distortion ratio (SNDR) of 58.92 dB and a spurious-free dynamic range (SFDR) of 82.23 dB. The ADC core consumes only 180 mW, resulting in a Schreier figure of merits (FoMs) of 156.4 dB.

Modern e-commerce platforms rely heavily on modeling diverse user feedback to provide personalized services.Consequently, multitask learning has become an integral part of their ranking systems.However, existing multi-task learning methods encounter two main challenges: some lack explicit modeling of task relationships, resulting in inferior performance, while others have limited applicability due to being computationally intensive, having scalability issues, or relying on strong assumptions.To address these limitations and better fit our real-world scenario, pre-rank in Shopee Search, we introduce in this paper ResFlow, a lightweight multi-task learning framework that enables efficient cross-task information sharing via residual connections between corresponding layers of task networks.Extensive experiments on datasets from various scenarios and modalities demonstrate its superior performance and adaptability over state-of-the-art methods.The online A/B tests in Shopee Search showcase its practical value in large-scale industrial applications, evidenced by a 1.29% increase in OPU (order-per-user) without additional system latency.ResFlow is now fully deployed in the pre-rank module of Shopee Search.To facilitate efficient online deployment, we propose a novel offline metric Weighted Recall@K, which aligns well with our online metric OPU, addressing the longstanding online-offline metric misalignment issue.Besides, we propose to fuse scores from the multiple tasks additively when ranking items, which outperforms traditional multiplicative fusion.

Since the development of artificial intelligence technology, E-Commerce has gradually become one of the world's largest commercial markets. Within this domain, sales events, which are based on sociological mechanisms, play a significant role. E-Commerce platforms frequently offer sales and promotions to encourage users to purchase items, leading to significant changes in live environments. Learning-To-Rank (LTR) is a crucial component of E-Commerce search and recommendations, and substantial efforts have been devoted to this area. However, existing methods often assume an independent and identically distributed data setting, which does not account for the evolving distribution of online systems beyond online finetuning strategies. This limitation can lead to inaccurate predictions of user behaviors during sales events, resulting in significant loss of revenue. In addition, models must readjust themselves once sales have concluded in order to eliminate any effects caused by the sales events, leading to further regret. To address these limitations, we introduce a long-term E-Commerce search data set specifically designed to incubate LTR algorithms during such sales events, with the objective of advancing the capabilities of E-Commerce search engines. Our investigation focuses on typical industry practices and aims to identify potential solutions to address these challenges.

In recent years, recommender systems have advanced rapidly, where embedding learning for users and items plays a critical role. A standard method learns a unique embedding vector for each user and item. However, such a method has two important limitations in real-world applications: 1) it is hard to learn embeddings that generalize well for users and items with rare interactions; and 2) it may incur unbearably high memory costs when the number of users and items scales up. Existing approaches either can only address one of the limitations or have flawed overall performances. In this paper, we propose Clustered Embedding Learning (CEL) as an integrated solution to these two problems. CEL is a plug-and-play embedding learning framework that can be combined with any differentiable feature interaction model. It is capable of achieving improved performance, especially for cold users and items, with reduced memory cost. CEL enables automatic and dynamic clustering of users and items in a top-down fashion, where clustered entities jointly learn a shared embedding. The accelerated version of CEL has an optimal time complexity, which supports efficient online updates. Theoretically, we prove the identifiability and the existence of a unique optimal number of clusters for CEL in the context of nonnegative matrix factorization. Empirically, we validate the effectiveness of CEL on three public datasets and one business dataset, showing its consistently superior performance against current state-of-the-art methods. In particular, when incorporating CEL into the business model, it brings an improvement of in AUC, which translates into a significant revenue gain; meanwhile, the size of the embedding table gets 2650 times smaller.1

Item representation learning is crucial for search and recommendation tasks in e-commerce. In e-commerce, the instances (e.g., items, users) in different domains are always related. Such instance relationship across domains contains useful local information for transfer learning. However, existing transfer learning based approaches did not leverage this knowledge. In this paper, we report on our experience designing and deploying Prior-Guided Transfer Learning (PGTL) to bridge this gap. It utilizes the instance relationship across domains to extract prior knowledge for the target domain and leverages it to guide the fine-grained transfer learning for e-commerce item representation learning tasks. Rather than directly transferring knowledge from the source domain to the target domain, the prior knowledge can serve as a bridge to link both domains and enhance knowledge transfer, especially when the domain distribution discrepancy is large. Since its deployment on the Taiwanese portal of Taobao in Aug 2020, PGTL has significantly improved the item exposure rate and item click-through rate compared to previous approaches

In the past decade, recommender systems for e-commerce have witnessed significant advancement. Recommendation scenarios can be divided into different type (e.g., pre-, during-, post-purchase, campaign, promotion, bundle) for different user groups or different businesses. For different scenarios, the goals of recommendation are different. This is reflected by the different performance metrics employed. In addition, online promotional campaigns, which attract high traffic volumes, are also a critical factor affecting e-commerce recommender systems. Typically, prior to a promotional campaign, the Add-to-Cart Rate (ACR) is the target of optimization. During the campaign, this changes to Gross Merchandise Volumes (GMV). Immediately after the campaign, it becomes Click Through Rates CTR. Dynamically adapting among these potentially conflicting optimization objectives is an important capability for recommender systems deployed in real-world e-commerce platforms. In this paper, we report our experience designing and deploying the Deep Controllable Learning-To-Rank (DC-LTR) recommender system to address this challenge. It enhances the feedback controller in LTR with multi-objective optimization so as to maximize different objectives under constraints. Its ability to dynamically adapt to changing business objectives has resulted in significant business advantages. Since September 2019, DC-LTR has become a core service enabling adaptive online training and real-time deployment ranking models based on changing business objectives in AliExpress and Lazada. Under both everyday use scenarios and peak load scenarios during large promotional campaigns, DC-LTR has achieved significant improvements in satisfying real-world business objectives.

Tasks such as search and recommendation have become increasingly important for E-commerce to deal with the information overload problem. To meet the diverse needs of different users, personalization plays an important role. In many large portals such as Taobao and Amazon, there are a bunch of different types of search and recommendation tasks operating simultaneously for personalization. However, most of current techniques address each task separately. This is suboptimal as no information about users shared across different tasks.

With the rapidly increasing number of personal image collections on the web, it is of great importance to annotate these user-uploaded images in personalized manner. But personalized image annotation is largely ignored by the mainstream of image annotation research. In this paper, we focus on personalizing the automatic image annotation by proposing a general framework which jointly exploits the generic content-based image annotation, personal image tagging history and the content of personal history images. In our framework, two sets of candidate annotations are extracted for each image based on content-based annotation and personal image tagging history. Considering that the user's interest may not stay the same, when exploiting the personal image tagging history, we also take the content of personal history images into account to avoid the noise. To get the final annotations, we propose an unsupervised algorithm based on reinforcement learning to combine the above two candidate annotation sets. Encouraging results show that the proposed framework is effective and promising for personalizing automatic image annotation.