Experiments on the LongLaMP dataset show that PrefReward outperforms non-personalized and retrieval-based baselines in both generation quality and personalization interpretability.
Abstract
Large Language Models (LLMs) have demonstrated remarkable ability in generating personalized content by leveraging user histories and contextual cues. However, most existing personalization approaches rely on implicit representations within model parameters, making it difficult to interpret user-specific preferences or effectively handle long-context dependencies. To address these challenges, we propose PrefReward, a novel preference-aware generative framework that explicitly models user styles through a structured preference matrix and integrates it into the decoding process as a reward signal. PrefReward consists of two stages: (1) extracting a user-specific preference matrix that summarizes individual stylistic tendencies, and (2) using the matrix to guide generation via a KL-divergence-based reward function. Experiments on the LongLaMP dataset show that PrefReward outperforms non-personalized and retrieval-based baselines in both generation quality and personalization interpretability.
Large language models (LLMs) are increasingly used in personalized applications, where responses must align with individual user preferences, histories, and profiles. A common approach is to inject user information into the prompt at inference time. However, existing methods typically rely on flat profile representations and similarity-based Top-K retrieval, which often introduce noisy or marginally relevant context under limited prompt budgets. In this paper, we propose PersMem, a user profile memory framework for LLM personalization that addresses three key questions: what user information to store, how to organize it, and how to use it effectively during generation. First, we construct a structured user profile memory by extracting entities and concepts from user profiles. Second, we organize the memory into a hierarchical structure that separates short-term interactions from long-term knowledge, enabling scalable and coarse-to-fine retrieval. Third, we introduce a learning-based memory gating mechanism that adaptively selects long-term memory units under prompt budget constraints and directly optimizes answer quality through evaluated signals. Extensive experiments on multiple personalization tasks demonstrate that our approach consistently improves personalization effectiveness while reducing prompt length, compared with existing prompt-based and retrieval-based baselines.
Yang-Xu Liao, Yongheng Deng, Tianyuan Jiang et al.· Proceedings of the 32nd ACM...· 0 citations
Personalized language models aim to adapt responses to individual users, whose preferences are often latent and revealed gradually through interaction. Existing training-free methods rely on stored histories or retrieved memories, but they often struggle to reconcile long- term preferences with short-term topic-specific needs. To address this issue, we propose HyperTrace, a training-free framework that formulates online personalization as latent preference tracing. HyperTrace maintains interpretable natural-language hypotheses over short-term intent and long-term preferences, and updates them through an SMC-style reweight process using an LLM-based surrogate choice model. By updating these hypotheses across turns and sessions, HyperTrace enables personalization without parameter updates. Experiments on PRISM and PersonaMem-v2 show that HyperTrace improves response alignment, preference prediction, and profile consistency over strong online baselines, demonstrating the effectiveness of tracing latent user preferences for robust personalization. Code and scripts are available in the repository: https://github.com/jiseshen/HyperTrace.
Jian-Zhi Shen, Ke-Yu Mao, Ming-Hao Shao et al.· 0 citations
Multimodal large language models (MLLMs) can convert multimodal item content into structured descriptions used as semantic features for recommendation. Conventional content-only generation, however, cannot use downstream user signals to determine which semantics should be emphasized. Recent user-conditioned methods incorporate these signals through user histories or profiles, but they require user information at inference and make generation user-dependent. In this paper, we introduce RecoReward, which instead uses behavior-derived rewards during training and preserves content-only inference. To instantiate this idea in live-stream recommendation, we treat historically engaged users as a proxy for future target users and use observational non-target users to estimate affinity shared broadly across users. The Recommender Affinity Score (RAS) contrasts these signals to provide user-selective feedback for reinforcement learning, allowing the learned policy to generate a single shared description without user inputs. In our offline benchmark, RecoReward-9B outperforms its Qwen3.5-9B baseline and all other evaluated models across seven recall metrics. Online A/B testing also shows performance gains. These results show that RecoReward trains the MLLM to produce item features that benefit downstream recommendation while retaining content-only serving.
Guohong Mu, Yue-Yang Liu, Jiangxia Cao et al.· arXiv.org· 0 citations
Natural language user preferences provide an interpretable interface for LLM personalization. However, universal preference summaries often contain information irrelevant to a particular downstream task. Directly supplying the full preference summary therefore wastes context capacity and introduces cross-task distraction, while manually designing task-specific preference views is difficult to scale. In this work, we study \emph{task-specific preference adaptation}: given a universal user preference summary and a downstream task, derive a task-conditioned representation that preserves sufficient decision-relevant evidence while removing redundant context. To this end, we propose \textsc{AlignXada}, a training-free meta-learning framework that induces reusable textual refinement policies for adapting universal preference summaries to task-specific ones. The refinement policy is iteratively optimized by a meta learner through verbal reinforcement learning. Across 13 tasks and three downstream models (39 task--model cells), \textsc{AlignXada} achieves an average gain of 3.82 points, improving 33 cells while retaining only 22.8\% of the original profile tokens and outperforming RAG in 36 cells. An extended faithfulness analysis further shows that the refined profiles remain largely grounded in the source preferences while preserving task-relevant personalization signals, suggesting that profile-side adaptation serves as a practical complement to universal memory construction for lifelong personalized agents.
Yuting Liu, Wei Wu, Jianzhe Zhao et al.· 0 citations
This work proposes Snap2Review, a novel framework that harmonizes cross-modal retrieval with fine-grained preference learning, and introduces Tri-DPO, which employs a stratified negative sampling strategy with varying difficulty levels to force the model to discern fine-grained user preferences.
Honggyan Xu, Lei Zou, De-Xiang Zhao et al.· Proceedings of the Thirty-Fi...· 0 citations
Personalizing language models (LMs) to individual user preferences is essential for aligning responses with diverse goals and backgrounds. Existing methods typically train a separate adapter for each user or learn a reward model whose scores depend on the user. Despite explicitly optimizing for each user, these methods must learn from limited observations and therefore suffer from data sparsity and poor generalization to unseen users and domains. In-context learning (ICL) and Context Steering (CoS) can instead provide more effective personalization by conditioning the base LM directly on user context and leveraging its pretrained capabilities without per-user training. Yet neither adapts the influence of that context across decoding steps: ICL leaves it uncontrolled, whereas CoS applies a fixed steering coefficient and requires two LM forward passes per step. We propose Cautious Context Steering (CCS), which adds a lightweight adapter to a frozen backbone LM to decide at each token whether and how strongly user context should affect generation. The adapter learns this behavior from an oracle context-conditioned LM and preserves the base LM when the context is not helpful. A single CCS adapter trained on only one dataset improves generation quality both in-domain and across four out-of-distribution personalization benchmarks, demonstrating robust generalization to new users and domains. CCS also avoids per-user fine-tuning and the additional context-conditioned forward pass required by CoS, substantially reducing inference cost.
Gihoon Kim, Jeyoung Lee, S. Woo et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
