Recommender systems aim to leverage user interaction signals to recommend items that users are likely to be interested in. Motivated by the success of Large Language Model (LLMs), generative recommendation (GR) has recently gained increasing attention, typically following a two-stage paradigm: supervised fine-tuning followed by preference alignment. However, aligning generative recommenders with users' personalized preferences remains challenging, as user feedback is inherently heterogeneous and uncertain. Different types of user interaction signals reflect varying levels of intent and should therefore be modeled differently. In this work, we propose Adaptive Mix Preference Optimization (AMPO), an adaptive alignment framework that mixes likelihood and preference objectives with self-calibrated, sample-wise confidence adjustment. AMPO introduces an adaptive target margin that leverages the model's own probability ratio to modulate optimization strength: confident pairs receive full margins that reinforce correct rankings, while uncertain pairs receive reduced margins that prevent overfitting to ambiguous signals. Additionally, AMPO incorporates negative log-likelihood regularization on preferred items to counteract likelihood displacement, a phenomenon where contrastive objectives cause preferred and non-preferred probabilities to collapse simultaneously. Such a design eliminates the need for a reference model, yielding up to 2.5x speedup and 35% memory reduction. Extensive experiments on public benchmarks and a large-scale industrial dataset demonstrate consistent improvements in ranking metrics. Online A/B tests on a major e-commerce platform further confirm statistically significant gains in click-through and conversion rates. The code is available at https://github.com/jumbo-q/ampo.

Novelty recommendation sustains long-term user engagement by exposing users to content that is both relevant and meaningfully different from their recent consumption. In large-scale e-commerce, this requires composing coherent yet non-redundant recommendation lists, a task fundamentally constrained by the relevance-diversity trade-off. Large language models (LLMs) offer a unified generative paradigm for inferring user intent and producing semantically coherent candidates, yet industrial deployment faces two critical challenges: (i) scarce supervision for modeling novelty transitions and diversity-aware list construction, and (ii) reward granularity mismatch, where standard RL assigns coarse sequence-level rewards that fail to capture item-level redundancy and complementarity. We present BALANCE, a hierarchical reasoning-and-generation framework that decomposes novelty recommendation into three structured stages: generating a Novelty Tag for exploration direction, refining an Interest Topic for intent specification, and constructing a Recommendation List for facet coverage. We address data scarcity through a self-reflection pipeline that synthesizes high-quality supervision by integrating real behavior logs with structured rationales. We resolve granularity mismatch through Sequence-Item Policy Optimization (SIPO), which jointly optimizes sequence- and item-level objectives via granularity-aware advantage fusion. Extensive offline experiments and online A/B test on the JD.com recommender system, validate the performance of our method, highlighting its superior novelty and diversity without compromising relevance.

Generative Retrieval (GR) offers a promising paradigm for recommendation through next-token prediction (NTP). However, scaling it to large-scale industrial systems introduces three challenges: (i) within a single request, the identical model inputs may produce inconsistent outputs due to the pagination request mechanism; (ii) the prohibitive cost of encoding long user behavior sequences with multi-token item representations based on semantic IDs, and (iii) aligning the generative policy with nuanced user preference signals. We present GenRec, a preference-oriented generative framework deployed on the JD App https://www.jd.com that addresses above challenges within a single decoder-only architecture. For training objective, we propose Page-wise NTP task, which supervises over an entire interaction page rather than each interacted item individually, providing denser gradient signal and resolving the one-to-many ambiguity of point-wise training. On the prefilling side, an asymmetric linear Token Merger compresses multi-token Semantic IDs in the prompt while preserving full-resolution decoding, reducing input length by ~2× with negligible accuracy loss. To further align outputs with user satisfaction, we introduce GRPO-SR, a reinforcement learning method that pairs Group Relative Policy Optimization with NLL regularization for training stability, and employs Hybrid Rewards combining a dense reward model with a relevance gate to mitigate reward hacking. In month-long online A/B tests serving production traffic, GenRec achieves 9.5% improvement in click count and 8.7% in transaction count over the existing pipeline.

State-of-the-art approaches for click-through rate (CTR) prediction in industry predominantly rely on transformer-based networks or their variants. However, as user behavior sequences become longer, employing self-attention networks for CTR prediction within a constrained inference time presents a significant challenge. To address this, mainstream methods adopt a classical two-stage paradigm: a General Search Unit (GSU) for quickly retrieving relevant items from long-term behaviors, and an Exact Search Unit (ESU) for applying effective Multi-Head Target Attention (MHTA) over the items selected by the GSU. These two-stage algorithms have certain limitations. Firstly, the GSU needs to retrieve different target subsequences for different target items, restricting the ESU to a suboptimal MHTA network rather than a more effective transformer-based network. Secondly, the GSU retrieves only a subset of items from the user's behavior sequence, ignoring the evolution of user interests and the interrelationships between different points of interest. To overcome these challenges, we propose a novel end-to-end hierarchical user long-term behavior modeling network for CTR prediction (HBM). Specifically, we employ the multi-interest routing layer to channel the user's long-term behavior to several aggregated interest clusters. Furthermore, we introduce a fine interest learning network that selects the top-k interests from the initial aggregated representations. Subsequently, we employ a transformer network to model the user's behavior sequence associated with these top-k interests in a detailed manner, while also capturing the inherent correlations between different user interests at a coarse level. Finally, we integrate the coarse and fine interests. Extensive experiments on two real-world datasets demonstrate the effectiveness of our proposed methods. In addition, an online A/B test on the JD recommendation platform shows promising improvements, with a 2.15% increase in CTR and a 0.98% increase in CVR, accompanied by lower online inference latency.

Multi-task learning (MTL) has been widely applied in online advertising systems. To address the negative transfer issue, recent optimization methods emphasized the gradient alignment of directions or magnitudes. Since prior studies have proven that the shared modules contain both general and specific knowledge, overemphasizing on gradient alignment may crowd out task-specific knowledge. In this paper, we propose a transference-driven approach CoGrad that adaptively maximizes knowledge transference via Coordinated Gradient modification. We explicitly quantify the transference as loss reduction from one task to another, and optimize it to derive an auxiliary gradient. By incorporating this gradient into original task gradients, the model automatically maximizes inter-task transfer and minimizes individual losses, leading to general and specific knowledge harmonization. Besides, we introduce an efficient approximation of the Hessian matrix, making CoGrad computationally efficient. Both offline and online experiments verify that CoGrad significantly outperforms previous methods.

Click-through rate (CTR) prediction is a fundamental technique in recommendation and advertising systems. Recent studies have proved that learning a unified model to serve multiple domains is effective to improve the overall performance. However, it is still challenging to improve generalization across domains under limited training data, and hard to deploy current solutions due to computational complexity. In this paper, we propose AdaSparse for multi-domain CTR prediction, which learns adaptively sparse structure for each domain, achieving better generalization across domains with lower computational cost. We introduce domain-aware neuron-level weighting factors to measure the importance of neurons, with that for each domain our model can prune redundant neurons to improve generalization. We further add flexible sparsity regularizations to control the sparsity ratio of learned structures. Offline and online experiments show that AdaSparse outperforms previous multi-domain CTR models significantly.

Current bundle generation studies focus on generating a combination of items to improve user experience. In real-world applications, there is also a great need to produce bundle creatives that consist of mixture types of objects (e.g., items, slogans and templates) for achieving better promotion effect. We study a new problem named bundle creative generation: for given users, the goal is to generate personalized bundle creatives that the users will be interested in. To take both quality and efficiency into account, we propose a contrastive non-autoregressive model that captures user preferences with ingenious decoding objective. Experiments on large-scale real-world datasets verify that our proposed model shows significant advantages in terms of creative quality and generation speed.