In industrial recommender systems, conversion rate (CVR) is often used for traffic allocation, but fails to fully reflect recommendation effectiveness as it does not account for refund rate (RFR). Thus, net conversion rate (NetCVR), the probability that a clicked item is purchased and not refunded, is proposed to better show true user satisfaction and business value. Unlike CVR, NetCVR prediction involves a more complex multi-stage cascaded delay feedback phenomenon. The two cascaded delays Click->Conversion and Conversion->Refund in NetCVR have opposite effects. Therefore, traditional CVR methods cannot be directly applied. At present, the lack of relevant open-source datasets and online continuous training schemes poses a challenge. To address these, we first introduce CAscadal Sequences of Conversion And Delayed rEfund (CASCADE), the first large-scale open dataset derived from Taobao app for online continuous NetCVR prediction. We further analyze CASCADE and derive three key insights: (1) NetCVR exhibits clear temporal patterns necessitating online continuous modeling; (2) Cascaded modeling CVR and RFR for NetCVR outperforms directly modeling NetCVR; and (3) delay time, which correlated with both CVR and RFR, is an important feature for NetCVR prediction. Based on these insights, we propose neT convErsion caScaded modeLing and debiAsing method (TESLA). This continuous method features a CVR-RFR cascaded architecture, stage-wise debiasing, and a delay-time-aware ranking loss for efficient NetCVR prediction. Experiments show that TESLA outperforms state-of-the-art methods on CASCADE, achieving an absolute improvement of 12.41% in RI-AUC and 14.94% in RI-PRAUC on NetCVR over the strongest baseline. We hope this work provides a new direction for online delayed feedback modeling in NetCVR prediction. Our code and dataset are available at https://github.com/alimama-tech/NetCVR.
论文检索
输入标题、作者或关键词，从 954 篇学术成果中精准定位
Large language models (LLMs) have emerged as a promising paradigm for recommender systems, due to their powerful capabilities in global knowledge integration and reasoning. However, LLMs are inherently prone to confirmation bias -- the tendency to favor information that reinforces users' existing views -- which leads to an overemphasis on previously shown viewpoints and ignores diverse user beliefs for recommendations. To address this issue, in this paper, we propose SCoTRec, a social chain-of-thought reasoning framework for recommendation. SCoTRec first constructs sentiment-aware user profiles by extracting sentiment terms from user reviews. It then incorporates users' social sentiment information into the social chain-of-thought reasoning units to improve recommendations. In particular, we categorize the social chain-of-thought into sentiment-based pathways and apply human evaluation operations -- backtracking, discarding, retaining, and aggregating -- to simulate nuanced sentiment cognition and interpersonal influence, effectively alleviating confirmation bias. Extensive experiments on four benchmark datasets demonstrate the effectiveness of SCoTRec in alleviating confirmation bias and improving recommendations.
Recent agent-based recommendation frameworks aim to simulate user behaviors by incorporating memory mechanisms and prompting strategies, but they struggle with hallucinating non-existent items and full-catalog ranking. Besides, a largely underexplored opportunity lies in leveraging LLMs' commonsense reasoning to capture user intent through substitute and complement relationships between items, which are usually implicit in datasets and difficult for traditional ID-based recommenders to capture. In this work, we propose a novel LLM-agent framework, AgentDR, which bridges LLM reasoning with scalable recommendation tools. Our approach delegates full-ranking tasks to traditional models while utilizing LLMs to (i) integrate multiple recommendation outputs based on personalized tool suitability and (ii) reason over substitute and complement relationships grounded in user history. This design mitigates hallucination, scales to large catalogs, and enhances recommendation relevance through relational reasoning. Through extensive experiments on three public grocery datasets, we show that our framework achieves superior full-ranking performance, yielding on average a twofold improvement over its underlying tools. We also introduce a new LLM-based evaluation metric that jointly measures semantic alignment and ranking correctness.
Multi-modal recommendation has become essential with the rapid expansion of online platforms such as e-commerce and video-sharing applications. In this work, we focus on the Multi-Modal New Item Recommendation (MMNIR) problem, where items with multi-modal content but newly introduced items lack interaction history. The MMNIR problem is particularly challenging in two aspects: (1) a large number of new items are created rapidly over time without any interaction data, (2) not all existing interactions are equally useful, and it is non-trivial to identify informative behaviors from users with similar preferences. However, previous methods fail to identify users with similar preferences and to capture informative behaviors from historical data. Furthermore, conventional models primarily rely on simple co-occurring signals, leading to spurious neighbors and neglecting the informative behaviors of truly similar users with consistent preferences. To fill this gap, we propose Joint Similar User Exploration and Informative Behavior Guidance (SuperG) for solving the MMNIR problem. SuperG first proposes a similar user exploration module to identify users with similar preferences to the target user. Then it incorporates an informative behavior mining module to retrieve informative behaviors from both the target user and similar users' histories to support new item recommendation. Finally, SuperG proposes a behavior-guided hybrid recommendation module to incorporate the retrieved behavioral signals to guide the recommendation of new items. Our empirical study on three real datasets demonstrates that SuperG outperforms the state-of-the-art models under the MMNIR setting.
Federated cross-domain recommendation (Federated CDR) aims to collaboratively learn personalized recommendation models across heterogeneous domains while preserving data privacy. Recently, large language model (LLM)-based recommendation models have demonstrated impressive performance by leveraging LLMs' strong reasoning capabilities and broad knowledge. However, adopting LLM-based recommendation models in Federated CDR scenarios introduces new challenges. First, there exists a risk of overfitting with domain-specific local adapters. The magnitudes of locally optimized parameter updates often vary across domains, causing biased aggregation and overfitting toward domain-specific distributions. Second, unlike traditional recommendation models (e.g., collaborative filtering, bipartite graph-based methods) that learn explicit and comparable user/item representations, LLMs encode knowledge implicitly through autoregressive text generation training. This poses additional challenges for effectively measuring the cross-domain similarities under heterogeneity. To address these challenges, we propose an LLM-based framework for federated cross-domain recommendation, FeDecider. Specifically, FeDecider tackles the challenge of scale-specific noise by disentangling each client's low-rank updates and sharing only their directional components. To handle the need for flexible and effective integration, each client further learns personalized weights that achieve the data-aware integration of updates from other domains. Extensive experiments across diverse datasets validate the effectiveness of our proposed FeDecider.
The persistent challenges of data sparsity and class imbalance have long limited the development of recommender systems. Fortunately, line graph theory offers a novel perspective to overcome these issues. By transforming the user-item interaction bipartite graph into a line graph, the problems of data sparsity and class imbalance are elegantly reformulated as those of insufficient labeled nodes and imbalanced label distribution in the line graph domain. This reformulation allows us to directly apply mature techniques from node classification and imbalanced graph learning to address these core challenges. Inspired by this insight, we propose a Line Graph Data Augmentation (LGDA) strategy, which features two distinct characteristics. Firstly, it is a plug-and-play module that resolves data sparsity and imbalance without modifying the underlying recommendation framework. Secondly, it employs a targeted augmentation and confidence filtering mechanism to generate high-quality, balanced augmented data. Extensive experiments on four real-world datasets validate that LGDA effectively alleviates data sparsity and class imbalance, leading to significant improvements in both recommendation performance and system robustness.
Watch time prediction (WTP) has emerged as a pivotal task in short video recommendation systems, designed to quantify user engagement through continuous interaction modeling. Predicting users' watch times on videos often encounters fundamental challenges, including wide value ranges and imbalanced data distributions, which can lead to significant estimation bias when directly applying regression techniques. Recent studies have attempted to address these issues by converting the continuous watch time estimation into an ordinal regression task. While these methods demonstrate partial effectiveness, they exhibit notable limitations: (1) The discretization process frequently relies on bucket partitioning, inherently reducing prediction flexibility and accuracy. (2) The interdependencies among different partition intervals remain underutilized, missing opportunities for effective error correction. Inspired by language modeling paradigms, we propose a novel Generative Regression (GR) framework that reformulates WTP as a sequence generation task. Our approach employs structural discretization to enable nearly lossless value reconstruction while maintaining prediction flexibility. Through carefully designed vocabulary construction and label encoding schemes, each watch time is bijectively mapped to a token sequence. To mitigate the training-inference discrepancy caused by teacher-forcing, we introduce a curriculum learning with embedding mixup strategy that gradually transitions from guided to free-generation modes. We test our models extensively on two public datasets, a large-scale offline industrial dataset, and an online A/B test on Kuaishou App with over 400 million daily active users (DAU) and GR consistently outperforms existing state-of-the-art approaches significantly. Our code is available at https://github.com/snailma0229/GR.git.
From Prediction to Understanding: Leveraging Reasoning in Large Language Model-based Recommendations
Recently, large language models (LLMs) have shown great promise in sequential recommendation. Existing methods typically transform users' historical interactions into textual sequences and then feed these sequences into LLMs to generate recommended items. However, since the output of LLMs is limited to recommendations, the models must analyze user interactions and infer user preferences implicitly. This implicit approach constrains the expressive reasoning abilities of LLMs. Moreover, as the output of LLMs consists solely of recommended items, the resulting recommendations lack explainability. To address these issues, we propose RE2, which enables LLMs to explicitly generate reasoning content before providing recommendations. Making the reasoning process explicit helps elicit the reasoning abilities of LLMs and simultaneously enhances the explainability of recommendation results. RE2 consists of three steps: (1) Reasoning Collection, which collects reasoning data by guiding LLMs to generate both reasoning content and recommended items for each interaction sequence. (2) Pattern Imitation, which leverages the collected data to train LLMs via supervised fine-tuning to imitate the pattern of first generating reasoning content and then providing recommendations. (3) Pattern Internalization, which further internalizes this reasoning-and-recommendation pattern and enhances both recommendation performance and the rationality of the reasoning content through reinforcement learning. RE2 can be implemented under both the self-distillation and teacher-distillation frameworks, without requiring external user metadata such as user reviews. Experimental results demonstrate the effectiveness of RE2 in improving recommendation performance and in generating high-quality reasoning content. Furthermore, we show that RE2 can mitigate popularity bias while maintaining recommendation accuracy to some extent. Our data and code are available at https://github.com/zhiyuanc2001/RE2.
Users increasingly interact with content across multiple domains, resulting in sequential behaviors marked by frequent and complex transitions. While Cross-Domain Sequential Recommendation (CDSR) models two-domain interactions, Multi-Domain Sequential Recommendation (MDSR) introduces significantly more domain transitions, compounded by challenges such as domain heterogeneity and imbalance. Existing approaches often overlook the intricacies of domain transitions, tend to overfit to dense domains while underfitting sparse ones, and struggle to scale effectively as the number of domains increases. We propose GMFlowRec, an efficient generative framework for MDSR that models domain-aware transition trajectories via Gaussian Mixture Flow Matching. GMFlowRec integrates: (1) a unified dual-masked Transformer to disentangle domain-invariant and domain-specific intents, (2) a Gaussian Mixture flow field to capture diverse behavioral patterns, and (3) a domain-aligned prior to support frequent and sparse transitions. Extensive experiments on JD and Amazon datasets show GMFlowRec achieves up to 44% NDCG@5 improvement over state-of-the-art baselines while maintaining efficiency with a single unified backbone. Our code and data are available here.
YouTube is the world's most widely used video platform, with over 70% of content viewed through algorithmic recommendations. While prior audits have examined polarization in YouTube's long-form video recommendations, the platform's fast-growing Shorts feature remains understudied. In this paper, we present the first large-scale audit comparing political content exposure and engagement dynamics across short-form and long-form videos on YouTube. We design a matched audit based on the insight that many news media organizations publish both short and long versions of the same content and collect 50,000 pairs of long-form and short-form video recommendations from both political and nonpolitcal seed videos. We analyze recommendations along several dimensions: the frequency of political recommendations, the diversity of retrieved videos, the engagement those videos receive, and finally, the partisan alignment between recommended videos and seed videos. Our results highlight fundamental differences between each algorithm, which we hope we can inform future research in analyzing the impact of YouTube recommendations.
Federated recommendation systems (FRSs) have recently gained widespread attention due to their ability to train collaborative recommendation models without exchanging raw user data. However, existing FRSs face a severe challenge of data sparsity, which manifests at both the user and item levels. First, user data sparsity: some users may only have a small number of interactions with items, struggling to adequately train the personalized user embedding locally. Second, item data sparsity: some items may only receive a small number of user ratings, causing the global model to lack knowledge about them. Considering these, we propose the Knowledge Enhanced Federated Recommendation System named as KE-FedRS, of which the core idea is to enhance the knowledge of users with few interactions and items with few ratings at both the local and global levels. Specifically, at the local level, we introduce an auxiliary user embedding and average and aggregate this auxiliary embedding across similar users, thereby enriching the knowledge of the local user embedding. At the global level, we propose a hybrid client selection strategy based on item embedding discrepancies, prioritizing clients that exhibit greater divergence in item embeddings from others, thus enhancing the knowledge of items with fewer interactions in the global model. We conduct comprehensive experiments on four real-world datasets, and the results show that the proposed method consistently outperforms baseline approaches in terms of HR@10 and NDCG@10.
Existing LLM-driven recommendation systems (RS) suffer from over-reliance on a single pre-trained model, which limits adaptability across diverse scenarios due to differences in large language models' strengths in semantics, knowledge, and reasoning. To address this issue, we propose AMLrec (Adaptive Multi-LLM Recommendation), a dynamic routing-based adaptive multi-LLM collaboration framework that unifies two dominant paradigms—LLM as Recommender and LLM + Recommender—through decision knowledge complementation. For each user or item, a lightweight encoder generates embeddings that are compared with learnable LLM prototypes using cosine similarity to select the most suitable models. In the first paradigm, selected LLMs generate recommendations via structured prompts, and their outputs are aggregated to form the final recommendation list. In the second paradigm, chosen LLMs produce semantic embeddings, which are fused with learnable embeddings after PCA-based dimensionality reduction and aligned using a lightweight adapter to bridge distribution gaps. Notably, AMLrec does not require fine-tuning of the underlying LLMs, significantly reducing computational overhead. Experiments on real-world datasets demonstrate that the proposed approach consistently outperforms single-LLM baselines across all evaluation metrics, validating its effectiveness. The main contributions of this work are threefold: introducing dynamic routing for multi-LLM recommendation system collaboration, proposing a unified architecture that harmonizes both paradigms, and enabling efficient adaptation without LLM fine-tuning. The code is available at https://github.com/Jiale-12138/AMLrec.
Customer Lifetime Value (LTV) prediction, a central problem in modern marketing, is characterized by a unique zero-inflated and long-tail data distribution. This distribution presents two fundamental challenges: (1) the vast majority of low-to-medium value users numerically overwhelm the small but critically important segment of high-value ''whale'' users, and (2) significant value heterogeneity exists even within the low-to-medium value user base. Common approaches either rely on rigid statistical assumptions or attempt to decouple ranking and regression using ordered buckets; however, they often enforce ordinality through loss-based constraints rather than inherent architectural design, failing to balance global accuracy with high-value precision. To address this gap, we propose Conditional Cascaded Ordinal-Residual Networks (CC-OR-Net), a novel unified framework that achieves a more robust decoupling through structural decomposition, where ranking is architecturally guaranteed. CC-OR-Net integrates three specialized components: a structural ordinal decomposition module for robust ranking, an intra-bucket residual module for fine-grained regression, and a targeted high-value augmentation module for precision on top-tier users. Evaluated on real-world datasets with over 300M users, CC-OR-Net achieves a superior trade-off across all key business metrics, outperforming state-of-the-art methods in creating a holistic and commercially valuable LTV prediction solution.
Search and Recommendation (S&R) are core information access channels on modern multi-scenario platforms. Existing joint S&R models face two critical challenges: (1) cross-scenario interest inconsistency, failing to unify explicit search intent (queries) and implicit recommendation intent (behavioral interactions) into coherent user interest representations; (2) severe S&R trade-off, where enhancing one task degrades the other due to static knowledge sharing and unbalanced feature utilization. To address these issues, we propose MinSAR , a novel framework focusing on cross-S&R user interest consistency. It integrates two key innovations: a Unified Interest Generation (UIG) module using Vector Quantized-Variational Autoencoder (VQ-VAE) to fuse long-term user preferences (via a user-specific memory network) and dynamic short-term contextual behaviors, generating compact cross-scenario latent representations that bridge explicit and implicit intents. Additionally, an Interest-Guided Attention Expert Network replaces static multi-task gating with intent-aware weight allocation. Guided by UIG's unified interest, it dynamically balances cross-S&R shared knowledge and task-specific expertise (semantic matching for search, collaborative filtering for recommendation), mitigating inter-task conflicts. Extensive experiments on two real-world datasets (KuaiSAR and Amazon Kindle Store) against 13 baselines show MinSAR outperforms state-of-the-art joint S&R models. Further analysis confirms its ability to eliminate the S&R performance trade-off.
Cross-domain sequential recommendation (CDSR) aims to utilize users' interactions across multiple domains to alleviate the problem of interaction sparsity that is prevalent in web platforms, thereby providing more accurate personalized recommendations. Although current CDSR methods have made some progress, they suffer from two main limitations: (i) assuming uniformly distributed interactions over time; and (ii) neglecting temporal influences during cross-domain transfer. In order to address the above issues, we propose a novel Time-Aware Cross-Domain Sequential Recommendation framework (TA-CDSR ). First, we design a time-sensitive attention which captures user preferences over time by decoupling interaction sequences and time sequences. Second, we propose a time-guided preference generator that can reconstruct the lacking interactions in the target domain by taking the source domain interactions time as guidance information. Finally, we design a multi-scale time windows based domain transfer module, which can dynamically identify the temporal interaction density and thus adaptively assign the weights of cross-domain information. Extensive experiments on three real-world datasets indicate that TA-CDSR achieves competitive time complexity while outperforming other baselines.
In recent years, substantial research has integrated multimodal item metadata into recommender systems, often by using pre?trained multimodal foundation models to encode such data. Since these models are not originally trained for recommendation tasks, recent works efficiently adapt them via parameter?efficient fine?tuning (PEFT). However, even with PEFT, item embeddings from multimodal foundation models remain user-blind: item embeddings are not conditioned on user interests, despite the fact that users with diverse interests attend to different item aspects. To address this limitation, we propose PerPEFT, a personalized PEFT strategy for multimodal recommendation. Specifically, PerPEFT groups users by interest and assigns a distinct PEFT module to each group, enabling each module to capture the fine?grained item aspects most predictive of that group's purchase decisions. We further introduce a specialized training technique that strengthens this user?group conditioning. Notably, PerPEFT is PEFT-agnostic and can be paired with any PEFT method applicable to multimodal foundation models. Through extensive experiments, we show that (1) PerPEFT outperforms the strongest baseline by up to 15.3% (NDCG@20) and (2) delivers consistent gains across diverse PEFT variants. It is noteworthy that, even with personalization, PEFT remains lightweight, adding only 1.3% of the parameter count of the foundation model. We provide our code and datasets at https://github.com/kswoo97/PerPEFT.
Personalized Learning Path Planning (PLPP) aims to design adaptive learning paths that align with individual goals. While large language models (LLMs) show potential in personalizing learning experiences, existing approaches often lack mechanisms for goal-aligned planning. We introduce Pxplore, a novel framework for PLPP that integrates a reinforcement-based training paradigm and an LLM-driven educational architecture. We design a structured learner state model and an automated reward function that transforms abstract objectives into computable signals. We train the policy combining supervised fine-tuning (SFT) and Group Relative Policy Optimization (GRPO), and deploy it within a real-world learning platform. Extensive experiments validate Pxplore's effectiveness in producing coherent, personalized, and goal-driven learning paths. We release our code and dataset at https://github.com/Pxplore/pxplore-algo.
Large Language Models (LLMs) have been widely adopted in ranking systems, specifically for reranking tasks. Despite the effectiveness, the auto-regressive decoding of LLMs leads to inference latency due to the memory-bandwidth-bound. To alleviate this bottleneck, prior studies have explored single token decoding as an approximation, but they suffer from performance degradation at the tail positions of ranking. In this paper, we propose a Single Forward Pass (SFP)-based method to pre-verify multiple rankings using tree attention, approximating auto-regressive decoding by relevant sub-rankings at each step. However, verifying all possible ranking permutations will lead to factorial-level token computation (N!), making it intractable within SFP. To this end, we first reduce item ranking permutations to combinations (2N), based on the empirical observation that the LLM's next-item generation is less sensitive to the exact ordering of preceding items. Furthermore, we divide the full ranking into K sub-rankings and aggregate their individual probabilities, which further decreases the verification space to 2⌈N/K⌉ • K << 2N. However, naïvely aggregating sub-ranking probabilities leads to inaccurate estimates in listwise ranking. To overcome this, we introduce a Möbius inversion model that explicitly decompose the individual contribution of subsets within a complete lattice, as verified by tree attention. Then, we learn their higher-order effects with a hierarchical self-attention model to reconstruct the full ranking probability. Experiments on both information retrieval and recommendation tasks show the effectiveness of our proposed method.
Live streaming platforms have recently gained popularity due to their immediacy and entertainment value, highlighting the need for streaming recommender systems that can adapt to the dynamic nature of evolving content, real-time interactions, and changing user interests. The ''live room'' plays a central role in modeling this dynamic environment, as it not only connects users with streamers but also serves as a key channel for collecting fine-grained user feedback. More specifically, the frequent interactions of users within a live room provide detailed dynamic collaboration information, reflecting the streamer's real-time topic and users' dynamic interests. However, existing studies have not thoroughly investigated the dynamics of room-level collaborative information. In this paper, we address this gap by emphasizing two perspectives: the evolving tripartite interaction information among rooms, streamers, and users, and the real-time intra-room collaboration information. We propose DCGLive, a Dynamic Collaboration-aware Graph learning approach for Live streaming recommendation. Specifically, we first construct two dynamic bipartite graphs to perceive the evolving tripartite interaction and generate real-time representations of streamers, rooms, and users. To account for the dynamic nature of live streaming, we design a set of non-parametric, collaboration-aware indicators that weight intra-room interactions based on both temporal recency and frequency, while guiding the embedding updating process for both users and rooms. Additionally, to address the cold-start challenge of newly created live rooms in real time, we propose a room representation initialization mechanism that balances both the relevance and the dynamics among different rooms hosted by the same streamer. Experiments conducted on both commercial and public datasets demonstrate that DCGLive consistently outperforms the baseline models. Our code is available at https://github.com/imgkkk574/DCGLive.
Graph-based collaborative filtering has advanced by modeling higher-order interactions, yet performance remains constrained by underlying geometric assumptions and propagation schemes. User-item interaction graphs typically exhibit pronounced topological heterogeneity, whereas existing methods rely on a fixed, homogeneous geometry and employ tangent space aggregation. To address these fundamental limitations, this paper introduces Adaptive Geometric Collaborative Filtering (AGCF), a novel method rooted in Hamiltonian dynamics, which reframes representation learning as a physical process evolving on a time-varying manifold. AGCF is distinguished by an integrated design comprising: (1) a learnable, node-dependent Riemannian metric that construct a continuous heterogeneous manifold aligned with local topology; (2) unified dynamic trajectories that achieve intrinsic propagation without tangent space approximations; (3) a channel-wise metric that captures semantic anisotropy in the feature space. We rigorously prove global existence and uniqueness of the induced dynamics and explain the mechanism enabling long-range information propagation. Extensive experiments on five benchmark datasets show consistent gains over representative baselines.