论文检索
输入标题、作者或关键词，从 100,903 篇学术成果中精准定位
Generative recommendation systems typically leverage Semantic Identifiers (SIDs), which represent each item as a sequence of tokens that encode semantic information. However, representing item ID with multiple SIDs significantly increases input sequence length, which is a major determinant of computational complexity and memory consumption. While existing efforts primarily focus on optimizing attention computation and KV cache, we propose RASTP (Representation-Aware Semantic Token Pruning), which directly prunes less informative tokens in the input sequence. Specifically, RASTP evaluates token importance by combining semantic saliency, measured via representation magnitude, and attention centrality, derived from cumulative attention weights. Since RASTP dynamically prunes low-information or irrelevant semantic tokens, experiments on three real-world Amazon datasets show that RASTP reduces training time by 26.7%, while maintaining or slightly improving recommendation performance. The code has been open-sourced at https://github.com/Yuzt-zju/RASTP.
Semantic IDs serve as a key component in generative recommendation systems. They not only incorporate open-world knowledge from large language models (LLMs) but also compress the semantic space to reduce generation difficulty. However, existing methods suffer from two major limitations: (1) the lack of contextual awareness in generation tasks leads to a gap between the Semantic ID codebook space and the generation space, resulting in suboptimal recommendations; and (2) suboptimal quantization methods exacerbate semantic loss in LLMs. To address these issues, we propose Dual-Flow Orthogonal Semantic IDs (DOS) method. Specifically, DOS employs a user–item dual-flow framework that leverages collaborative signals to align the Semantic ID codebook space with the generation space. Furthermore, we introduce an orthogonal residual quantization scheme that rotates the semantic space to an appropriate orientation, thereby maximizing semantic preservation. Extensive offline experiments and online A/B testing demonstrate the effectiveness of DOS. The proposed method has been successfully deployed in Meituan's mobile application, serving hundreds of millions of users.
Multilingual Text-Centric Visual Question Answering (TEC-VQA) has become crucial for real-world applications, as it requires fine-grained understanding and reasoning over multilingual scene text. Recent advances in vision-language models (VLMs) have demonstrated strong potential in tackling multimodal tasks. However, most existing approaches rely primarily on textual Chain-of-Thought (CoT) and provide limited support for multilingual multimodal reasoning. To address this gap, we introduce LaV-CoT, the first Language-aware Visual CoT framework with Multi-Aspect Reward Optimization. LaV-CoT incorporates an interpretable multi-stage reasoning pipeline consisting of text summary with bounding box, language identification, spatial object-level captioning, and step-by-step logical reasoning. To improve reasoning accuracy and cross-lingual generalization, we propose a novel verifiable Multi-Aspect Reward Optimization in addition to supervised fine-tuning that incorporates rewards for linguistic consistency, structural fidelity, and response accuracy. Extensive evaluations on public datasets, including MMMB, Multilingual MMBench, and MTVQA, show that LaV-CoT outperforms open-source models of similar size by up to ~9.5% accuracy, even surpassing open-source models more than twice its size, and further exceeding several state-of-the-art proprietary models. Moreover, LaV-CoT has been integrated into our online Intelligent Document Processing platform. A further online A/B test demonstrates an \(\sim\)8.7% improvement in acceptance rate, validating its effectiveness in industrial deployment and commercial applications. Our code is available at this https://github.com/HJNVR/LaV-CoT repository.
The widespread use of Multimedia Messaging Service (MMS) has led to a significant increase in the circulation of malicious visual content, presenting new challenges for scalable content moderation systems. In this work, we address the problem of visual spam detection in MMS by introducing a domain-specific taxonomy of inappropriate image categories. Based on this taxonomy, we construct a balanced training dataset from publicly available image collections, and two additional evaluation benchmarks derived from real-world MMS messages, which to the best of our knowledge are not covered by existing public datasets. All datasets were verified and manually labeled in order to ensure high annotation quality in line with our taxonomy. Furthermore, we show how to efficiently classify across eight categories related to MMS spam using an adapted CLIP-based architecture. Our empirical evaluation demonstrates that a fine-tuned CLIP model achieves strong accuracy that closely matches the performance of GPT-4o, but at a significantly lower cost which is crucial when performing at scale.
Balancing multiple objectives is critical for user satisfaction in modern recommender and search systems, yet current Multi-Task Fusion (MTF) methods rely on static, manually-tuned weights that fail to capture individual user intent. While Reinforcement Learning (RL) offers a path to personalization, traditional approaches often falter due to training instability and the sparse rewards inherent in these large-scale systems. To address these limitations, we propose Group-Relative Reinforcement Learning with Adaptive Dirichlet Exploration (GRADE), a novel and robust framework for personalized multi-task fusion. GRADE leverages a critic-free, Group Relative Policy Optimization (GRPO) paradigm, enabling stable and efficient policy learning by evaluating the relative performance of candidate weight groups. Its core innovations include employing the Dirichlet distribution for principled and structured exploration of the weight space, and a composite reward function that combines sparse user feedback with dense model priors and rule-based constraints to guide the search effectively. Deployed in the in-app marketplace of an application with over hundreds of millions daily active users, GRADE significantly outperforms established baselines, achieving substantial gains in rigorous large-scale A/B tests: +0.595% in CTR, +1.193% in CVR, +1.788% in OPM, and +1.568% in total order volume. Following its strong performance, GRADE has been fully deployed in the marketplace search scenario of Kuaishou, serving hundreds of millions of users.
Query-service relevance prediction in e-commerce search systems faces strict latency requirements that prevent the direct application of Large Language Models (LLMs). To bridge this gap, we propose a two-stage reasoning distillation framework to transfer reasoning capabilities from a powerful teacher LLM to a lightweight, deployment-friendly student model. In the first stage, we address the limitations of general-purpose LLMs by constructing a domain-adapted teacher model. This is achieved through a three-step process: domain-adaptive pre-training to inject platform knowledge, supervised fine-tuning to elicit reasoning skills, and preference optimization with a multi-dimensional reward model to ensure the generation of reliable and preference-aligned reasoning paths. This teacher can then automatically annotate massive query-service pairs from search logs with both relevance labels and reasoning chains. In the second stage, to address the challenges of architectural heterogeneity in standard distillation, we introduce Contrastive Reasoning Self-Distillation (CRSD). By modeling the behavior of the same student model under ''standard'' and ''reasoning-augmented'' inputs as a teacher-student relationship, CRSD enables the lightweight model to internalize the teacher's complex decision-making mechanisms without needing the explicit reasoning path at inference. Offline evaluations and online A/B testing in the Meituan search advertising system demonstrate that our framework achieves significant improvements across multiple metrics, validating its effectiveness and practical value.
Kuaishou serving hundreds of millions of searches daily, the quality of short-video search is paramount. However, it suffers from a severe Matthew effect on long-tail queries: sparse user behavior data causes models to amplify low-quality content such as clickbait and shallow content. The recent advancements in Large Language Models (LLMs) offer a new paradigm, as their inherent world knowledge provides a powerful mechanism to assess content quality, agnostic to sparse user interactions. To this end, we propose a LLM-driven multimodal reranking framework, which estimates user experience without real user behavior. The approach involves a two-stage training process: the first stage uses multimodal evidence to construct high-quality annotations for supervised fine-tuning, while the second stage incorporates pairwise preference optimization to help the model learn partial orderings among candidates. At inference time, the resulting experience scores are used to promote high-quality but underexposed videos in reranking, and further guide page-level optimization through reinforcement learning. Experiments show that the proposed method achieves consistent improvements over strong baselines in offline metrics including AUC, NDCG@K, and human preference judgement. An online A/B test covering 15% of traffic further demonstrates gains in both user experience and consumption metrics, confirming the practical value of the approach in long-tail video search scenarios.
The proliferation of ride-hailing services has fundamentally transformed urban mobility patterns, making accurate ride-hailing forecasting crucial for optimizing passenger experience and urban transportation efficiency. However, ride-hailing forecasting faces significant challenges due to geospatial heterogeneity and high susceptibility to external events. This paper proposes MVGR-Net (Multi-View Geospatial Representation Learning), a novel framework that addresses these challenges through a two-stage approach. In the pre-training stage, we learn comprehensive geospatial representations by integrating Points-of-Interest and temporal mobility patterns to capture regional characteristics from both semantic attribute and temporal mobility pattern views. The forecasting stage leverages these representations through a prompt-empowered framework that fine-tunes Large Language Models while incorporating external events. Extensive experiments on DiDi's real-world datasets demonstrate the state-of-the-art performance.
Wikipedia is a critical source of information for millions of users across the Web. It serves as a key resource for large language models, search engines, question-answering systems, and other Web-based applications. In Wikipedia, content needs to be verifiable, meaning that readers can check that claims are backed by references to reliable sources. This depends on manual verification by editors, an effective but labor-intensive process, especially given the high volume of daily edits. To address this challenge, we introduce a multilingual machine learning system to assist editors in identifying claims requiring citations. Our approach is tested in 10 language editions of Wikipedia, outperforming existing benchmarks for reference need assessment. We not only consider machine learning evaluation metrics but also system requirements, allowing us to explore the trade-offs between model accuracy and computational efficiency under real-world infrastructure constraints. We deploy our system in production and release data and code to support further research.
Effective relevance modeling is crucial for e-commerce search, as it aligns search results with user intent and enhances customer experience. Recent work has leveraged large language models (LLMs) to address the limitations of traditional relevance models, particularly their inability to handle long-tail and ambiguous queries. By incorporating Chain-of-Thought (CoT) reasoning, these approaches further improve both accuracy and interpretability through explicit, multi-step reasoning pathways. However, two key limitations remain: (1) most existing approaches rely on single-perspective CoT reasoning, which fails to capture the multifaceted nature of e-commerce relevance (e.g., user intent vs. attribute-level matching vs. business-specific rules); and (2) although CoT-enhanced LLMs offer rich reasoning capabilities, their high inference latency necessitates knowledge distillation for real-time deployment, yet current distillation methods discard the CoT rationale structure at inference, using it only as a transient auxiliary signal and thereby forfeiting its reasoning utility for online serving. To address these challenges, we propose a novel framework that better exploits CoT semantics throughout the optimization pipeline. Specifically, the teacher model leverages Multi-Perspective CoT (MPCoT) to generate diverse rationales and combines Supervised Fine-Tuning (SFT) with Direct Preference Optimization (DPO) to construct a more robust reasoner. For distillation, we introduce Latent Reasoning Knowledge Distillation (LRKD), which endows a student model with a lightweight inference-time latent reasoning extractor, allowing efficient and low-latency internalization of the LLM's sophisticated reasoning capabilities. Evaluated through offline experiments and online A/B tests on an e-commerce search advertising platform serving tens of millions of users daily, our method delivers significant offline gains, along with a 1.42% online improvement in Revenue Per Mille (RPM) and a 0.4% increase in relevance satisfaction score (RS), demonstrating clear benefits in both commercial performance and user experience.
In recommendation systems, scaling up feature-interaction modules (e.g., Wukong, RankMixer) or user-behavior sequence modules (e.g., LONGER) has achieved notable success. However, these efforts typically proceed on separate tracks, which not only hinders bidirectional information exchange but also prevents unified optimization and scaling. In this paper, we propose Model, a unified Transformer backbone that simultaneously performs user-behavior sequence modeling and feature interaction. Model employs a unified tokenizer to convert both sequential and non-sequential attributes into a single token sequence. The stacked Model blocks share parameters across similar sequential tokens while assigning token-specific parameters to non-sequential tokens. Through causal attention and cross-request KV caching, Model enables precomputation and caching of intermediate representations, significantly reducing computational costs during both training and inference. Experimental results on industrial-scale datasets demonstrate that Model scales efficiently with increasing parameters, consistently outperforms strong baselines, and yields a 5.68% lift in per-user GMV in online A/B tests.
Autonomous Feature Engineering (AFE) is critical for improving predictive performance on tabular data by relieving humans from manual feature crafting. However, traditional AFE lacks the semantic guidance needed to fully exploit domain knowledge. Although large language models (LLMs) can, in principle, emulate experts, existing approaches typically operate in an open code space that directly generates and rewrites entire features; without a compositional structural representation and invariant constraints, edits are coarse and non-local, making it hard to distill interpretable features with high information content and rich hierarchical structure. We propose AFE-Master, a novel LLM-driven AFE framework that shifts feature construction from black-box evolution to structural and semantically rigorous search. AFE-Master uses a domain-specific language (DSL) to explicitly represent feature transformations and parses them into abstract syntax trees (ASTs), enabling the LLM to understand and manipulate feature structures with clear semantics. On this interpretable representation, we employ guided local search (GLS) over syntactic and semantic neighborhoods, making small, checkable edits that efficiently and controllably discover information-dense, hierarchically structured features. Experiments spanning Kaggle and OpenML benchmarks as well as multiple tabular models (XGBoost, MLP, and the frontier TabPFN) show significant gains. At industrial scale, we further conduct a large online A/B test on the advertising recommendation service of a major mobile app store. Starting from a mature, large feature set—167 expert-crafted features refined over two years—we add 20 features automatically generated by AFE-Master. On 100M+ live samples, a well-engineered FiBiNET model achieves +15.11% CPM and +3.01% CTR, demonstrating practical value and transferability under both massive sample volume and a high-feature-count production setting. These results indicate that AFE-Master's semantically guided approach can discover expert-level features beyond the reach of prior methods, pointing to a new generation of interpretable, high-performance AFE techniques.
Customer service automation has seen growing demand within digital transformation. Existing approaches either rely on modular system designs with extensive agent orchestration or employ over-simplified instruction schemas, providing limited guidance and poor generalizability. This paper introduces an orchestration-free framework using Task-Oriented Flowcharts (TOFs) to enable end-to-end automation without manual intervention. We first define the components and evaluation metrics for TOFs, then formalize a cost-efficient flowchart construction algorithm to abstract procedural knowledge from service dialogues. We emphasize local deployment of small language models and propose decentralized distillation with flowcharts to mitigate data scarcity and privacy issues in model training. Extensive experiments validate the effectiveness in various service tasks, with superior quantitative and application performance compared to strong baselines and market products. By releasing a web-based system demonstration with case studies, we aim to promote streamlined creation of future service automation.
Generative inference with Large Language Models (LLMs) is the cornerstone of web-scale AI search, where queries are answered using vast, heterogeneous documents retrieved via Retrieval-Augmented Generation (RAG). This paradigm is critically bottlenecked by the cost of self-attention mechanism on long context. The sheer diversity of retrieved web content (multi-sourced, multi-lingual, multi-faceted) makes simple Key-Value (KV) cache optimizations with pre-fixed subsets ineffective, demanding a dynamic, content-aware approach. This challenge, however, introduces a classic chicken-and-egg problem: the model cannot foresee the necessary KV entries for attention without first inferring on the content, yet doing so on the full context is prohibitively expensive. This paper introduces P&F, a unified framework that resolves this dilemma through a core ''probe-and-fetch'' mechanism, which ingeniously integrates with speculative decoding -- an acceleration approach already adopted in web-scale AI search. The probe step repurposes the speculative draft model: while generating candidate tokens, it simultaneously probes the context to predict the most salient KV entries the large model will need for attention. The fetch step immediately acts on this prediction, asynchronously fetching these sparse entries from memory. This synergistic design piggybacks the probing step onto the drafting process, allowing the expensive gathering of a sparse KV cache to be fully masked. Crucially, this co-design breaks the sequential dependency bottleneck that cripples naive integrations of speculative decoding and prefetching due to synchronization issues. Extensive experiments show P&F significantly outperforms state-of-the-art methods in throughput and scalability, offering a practical, drop-in solution. Extensive offline evaluations across various settings and datasets demonstrate that P&F yields superior throughput and scalability compared to advanced baselines, while maintaining model quality across diverse models and scales. In online settings, P&F delivers substantial gains in throughput improvements while preserving response quality, making it well-suited for large-scale industrial deployment in real-time AI Search services.