Amap's Spacetime-GR introduces a spatiotemporal-aware generative framework for industrial-scale POI recommendation, using hierarchical POI encoding, explicit spatiotemporal tokenization, and a three-stage training pipeline to achieve significant CTR and CVR gains in online deployment at ICDM 2026.
This paper introduces MDGR, a masked diffusion framework for generative recommendation that replaces autoregressive left-to-right decoding with a parallel mask-denoising process, achieving up to 6.56% relative improvement offline and significant online gains in Alibaba's advertising platform.
Alibaba International Intelligent Technology presented ten CIKM 2026 papers that introduce multimodal large‑model techniques—such as SAM‑D2Q, C2P, PRO‑Bid, UTTSI, GRC, CDNet, CAIM and SORT—to overcome bottlenecks in e‑commerce search, recommendation, auto‑bidding and CTR prediction, delivering substantial offline metric lifts and significant online gains in CTR, CVR, GMV and revenue.
The paper identifies a structural mismatch between probability‑driven beam search and reward‑driven RL fine‑tuning in generative recommendation, proposes V-STAR with value‑guided efficient decoding (VED) and sibling‑wise GRPO to align decoding and optimization, and demonstrates superior offline and online performance through extensive experiments and ablations.
The paper introduces S‑GRec, a semantic‑aware generative recommendation framework that decouples a lightweight online generator from an offline LLM‑based personalized semantic judge, using a novel asymmetric advantage policy optimization to align deep semantic understanding with commercial metrics without adding online latency.
The article recaps Kuaishou's June 13 technical salon, detailing the systematic evolution of generative recommendation—from scaling models to reasoning enhancements—through core projects OneReason, Pool‑Rec, OneSearch V2, and GR4AD, and announces the industry‑wide LLM‑Rec challenge for students.
This article introduces LWGR, a framework that uses personalized soft prompts generated by parallel codebooks and Lagrangian‑constrained knowledge fusion to integrate large language model world knowledge into generative recommendation, overcoming fixed‑prompt limitations and knowledge‑behavior conflicts, and demonstrates superior performance on public and industrial datasets with notable revenue gains in online A/B tests.
Alibaba International Intelligent Technology showcases nine industrial‑grade recommendation and search papers accepted at SIGIR, WWW, and WSDM 2026, detailing sparse scaling, counterfactual multi‑task learning, generative recommendation, MoE routing, and multimodal semantic ID breakthroughs with extensive offline and online results.
GR4AD, Kuaishou’s generative recommendation system, redesigns the entire ad pipeline—from tokenizing multimodal ad material to value‑aware learning, lazy decoding, and dynamic beam search—delivering over 4 % revenue lift, higher eCPM, and sub‑100 ms latency for more than 400 million users.
This paper introduces the TCA4Rec framework that directly aligns item‑level collaborative‑filtering preferences with token‑level objectives of large language models, presenting novel modules, extensive experiments, and analysis that demonstrate significant performance gains in generative recommendation tasks.
This article details JD Retail’s 9N‑LLM unified training engine, explaining the background of generative recommendation, the challenges of massive sparse and dense parameters, and the multi‑framework, multi‑hardware solutions—including efficient sample processing, large‑scale sparse embedding, dense scaling, UniAttention acceleration, and reinforcement‑learning integration—that enable industrial‑scale deployment.
This article details the OnePiece framework, which integrates context engineering, anchor item sequences, and progressive implicit reasoning into generative recommendation systems, achieving significant offline and online performance gains on Shopee Search by enhancing model inference, personalization, and computational efficiency.
This article surveys the evolution of generative recommendation systems powered by large language models, detailing their technical foundations, engineering challenges, recent breakthroughs, and future research directions, while highlighting why the paradigm shift is occurring now.
Amap upgrades its homepage recommendation by integrating large‑model capabilities—G‑Where, G‑Action, and G‑Plan—through semantic ID generation, item tokenization, and multi‑stage LLM training, achieving significant offline and online performance gains while illustrating a scalable generative recommendation framework.
The article presents the OnePiece framework, which injects LLM‑style context engineering and latent reasoning into item‑ID based search‑and‑recommendation models, details the design choices, training tricks, attention analysis, and reports online gains of around 1% GMV and ad revenue, offering a thorough technical dissection of generative recommendation in industrial settings.
This article explains why Meituan introduced a generative recommendation model, describes the MTGR architecture, data organization, training and inference engines built on TorchRec and TensorRT, reports performance gains and cost reductions, and outlines future directions such as simplifying the recommendation funnel and cross‑business heterogeneous modeling.
The paper identifies a “sandglass” bottleneck in Residual Quantization Semantic Identifiers, where middle‑layer tokens dominate, causing sparse paths and long‑tail distributions that hurt e‑commerce search performance, and demonstrates that adaptive pruning of these tokens restores accuracy and efficiency better than removing the layer entirely.
COBRA, Baidu’s new generative retrieval framework, unifies sparse ID generation and dense vector encoding through a cascaded architecture that first predicts hierarchical IDs then refines them into dense representations, achieving state‑of‑the‑art recall, NDCG and conversion gains across public benchmarks and large‑scale advertising production.
This article surveys recent advances in generative recommendation for CPS advertising, detailing explicit intent‑aware controllable product recommendation, multi‑objective optimization techniques based on reward‑in‑context and DPO, and the scalable One4All framework that unifies behavior and language modeling across diverse ad scenarios.
Since large language models have excelled in NLP, researchers are now enhancing CPS advertising recommendation systems by integrating generative LLMs for explicit intent perception, multi‑objective optimization, and a unified One4All framework, achieving significant offline and online performance gains across click‑through, conversion, and revenue metrics.
The paper presents a generative recommendation framework for JD Alliance advertising that combines semantic‑ID modeling, large‑model pre‑training and fine‑tuning, and Direct Preference Optimization (including Softmax‑DPO and β‑DPO) to jointly boost click‑through and conversion rates, achieving +0.6% UCTR and +8% UCVR in online tests while outlining future multi‑objective extensions.
This article explains how large language models reshape recommendation systems by simplifying pipelines, integrating world knowledge, and leveraging scaling laws, and details the engineering steps for deploying generative recall models—including product encoding, user prompting, model training, TensorRT‑LLM optimization, and continuous performance improvements.
Faced with overloaded job listings and low offer rates, a group of students built “WenZhi,” an AI‑driven job‑matching app that leverages Baidu’s ERNIE SDK, generative recommendation, and workflow orchestration to deliver personalized role suggestions and interview advice within minutes.
This article surveys how large language models reshape recommendation systems, presents a generative RS framework tailored for JD Alliance advertising, details material representation, model input, training and inference pipelines, and reports extensive offline and online experiments demonstrating its effectiveness on sparse user data.
This article surveys how large language models reshape recommendation systems, details a generative recommender framework for JD Alliance ads—including item representation, model input, training, and inference—presents extensive offline and online experiments, and discusses future optimization directions.
This article reviews recent advances in generative recommendation, highlighting challenges such as item representation and multimodal fusion, and summarizing four key research papers that propose novel tokenization, collaborative integration, and transformer-based multimodal approaches to improve recommendation performance.
