Short-video platforms have rapidly become a new generation of information retrieval systems, where users formulate queries to access desired videos. However, user queries, especially long-tail ones, often suffer from spelling errors, incomplete phrasing, and ambiguous intent, resulting in mismatches between user expectations and retrieved results. While large language models (LLMs) have shown success in long-tail query rewriting within e-commerce, they struggle on short-video platforms, where proprietary content such as short videos, live streams, micro dramas, and user social networks falls outside their training distribution. To address this challenge, we introduce CardRewriter, an LLM-based framework that incorporates domain-specific knowledge to enhance long-tail query rewriting. For each query, our method aggregates multi-source knowledge relevant to the query and summarizes it into an informative and query-relevant knowledge card. This card then guides the LLM to better capture user intent and produce more effective query rewrites. We optimize CardRewriter using a two-stage training pipeline: supervised fine-tuning followed by group relative policy optimization, with a tailored reward system balancing query relevance and retrieval effectiveness. Offline experiments show that CardRewriter substantially improves rewriting quality for queries targeting proprietary content. Online A/B testing further confirms significant gains in long-view rate (LVR) and click-through rate (CTR), along with a notable reduction in initiative query reformulation rate (IQRR). Since September 2025, CardRewriter has been deployed on Kuaishou, one of China's largest short-video platforms, serving hundreds of millions of users daily.

In a multi-stage recommendation system, reranking plays a crucial role in modeling intra-list correlations among items. A key challenge lies in exploring optimal sequences within the combinatorial space of permutations. Recent research follows a two-stage (generator-evaluator) paradigm, where a generator produces multiple feasible sequences, and an evaluator selects the best one. In practice, the generator is typically implemented as an autoregressive model. However, these two-stage methods face two main challenges. First, the separation of the generator and evaluator hinders end-to-end training. Second, autoregressive generators suffer from inference efficiency. In this work, we propose a Unified Generative Efficient Reranking Framework (GReF) to address the two primary challenges. Specifically, we introduce Gen-Reranker, an autoregressive generator featuring a bidirectional encoder and a dynamic autoregressive decoder to generate causal reranking sequences. Subsequently, we pre-train Gen-Reranker on the item exposure order for high-quality parameter initialization. To eliminate the need for the evaluator while integrating sequence-level evaluation during training for end-to-end optimization, we propose post-training the model through Rerank-DPO. Moreover, for efficient autoregressive inference, we introduce ordered multi-token prediction (OMTP), which trains Gen-Reranker to simultaneously generate multiple future items while preserving their order, ensuring practical deployment in real-time recommender systems. Extensive offline experiments demonstrate that GReF outperforms state-of-the-art reranking methods while achieving latency that is nearly comparable to non-autoregressive models. Additionally, GReF has also been deployed in a real-world video app Kuaishou with over 300 million daily active users, significantly improving online recommendation quality.

Retrieval aims to find the top-k items most relevant to a query/user from a large dataset. Traditional retrieval models represent queries/users and items as embedding vectors and use Approximate Nearest Neighbor (ANN) search for retrieval. Recently, researchers have proposed a generative-based retrieval method that represents items as token sequences and uses a decoder model for autoregressive training. Compared to traditional methods, this approach uses more complex models and integrates index structure during training, leading to better performance. However, these methods remain two-stage processes, where index construction is separate from the retrieval model, limiting the model's overall capacity. Additionally, existing methods construct indices by clustering pre-trained item representations in Euclidean space. However, real-world scenarios are more complex, making this approach less accurate. To address these issues, we propose a \underline{U}nified framework for \underline{R}etrieval and \underline{I}ndexing, termed \textbf{URI}. URI ensures strong consistency between index construction and the retrieval model, typically a Transformer decoder. URI simultaneously builds the index and trains the decoder, constructing the index through the decoder itself. It no longer relies on one-sided item representations in Euclidean space but constructs the index within the interactive space between queries and items. Experimental comparisons on three real-world datasets show that URI significantly outperforms existing methods.

Industrial recommendation systems (RS) rely on the multi-stage pipeline to balance effectiveness and efficiency when delivering items from a vast corpus to users. Existing RS benchmark datasets primarily focus on the exposure space, where novel RS algorithms are trained and evaluated. However, when these algorithms transition to real-world industrial RS, they face two critical challenges: (1) handling unexposed items—a significantly larger space than the exposed one, profoundly impacting their practical performance; and (2) overlooking the intricate interplay between multiple stages of the recommendation pipeline, resulting in suboptimal system performance. To bridge the gap between offline RS benchmarks and real-world online environments, we introduce RecFlow—an industrial full-flow recommendation dataset. Unlike existing datasets, RecFlow includes samples not only from the exposure space but also from unexposed items filtered at each stage of the RS funnel. RecFlow comprises 38 million interactions from 42,000 users across nearly 9 million items with additional 1.9 billion stage samples collected from 9.3 million online requests over 37 days and spanning 6 stages. Leveraging RecFlow, we conduct extensive experiments to demonstrate its potential in designing novel algorithms that enhance effectiveness by incorporating stage-specific samples. Some of these algorithms have already been deployed online at KuaiShou, consistently yielding significant gains. We propose RecFlow as the first comprehensive whole-pipeline benchmark dataset for the RS community, enabling research on algorithm design across the entire recommendation pipeline, including selection bias study, debiased algorithms, multi-stage consistency and optimality, multi-task recommendation, and user behavior modeling.

Ranking models primarily focus on modeling the relative order of predictions while often neglecting the significance of the accuracy of their absolute values. However, accurate absolute values are essential for certain downstream tasks, necessitating the calibration of the original predictions. To address this, existing calibration approaches typically employ predefined transformation functions with order-preserving properties to adjust the original predictions. Unfortunately, these functions often adhere to fixed forms, such as piece-wise linear functions, which exhibit limited expressiveness and flexibility, thereby constraining their effectiveness in complex calibration scenarios. To mitigate this issue, we propose implementing a calibrator using an Unconstrained Monotonic Neural Network (UMNN), which can learn arbitrary monotonic functions with great modeling power. This approach significantly relaxes the constraints on the calibrator, improving its flexibility and expressiveness while avoiding excessively distorting the original predictions by requiring monotonicity. Furthermore, to optimize this highly flexible network for calibration, we introduce a novel additional loss function termed Smooth Calibration Loss (SCLoss), which aims to fulfill a necessary condition for achieving the ideal calibration state. Extensive offline experiments confirm the effectiveness of our method in achieving superior calibration performance. Moreover, deployment in Kuaishou's large-scale online video ranking system demonstrates that the method's calibration improvements translate into enhanced business metrics. The source code is available at https://github.com/baiyimeng/UMC.

Sequential Recommendation (SR) plays a pivotal role in recommender systems by tailoring recommendations to user preferences based on their non-stationary historical interactions. Achieving high-quality performance in SR requires attention to both item representation and diversity. However, designing an SR method that simultaneously optimizes these merits remains a long-standing challenge. In this study, we address this issue by integrating recent generative Diffusion Models (DM) into SR. DM has demonstrated utility in representation learning and diverse image generation. Nevertheless, a straightforward combination of SR and DM leads to sub-optimal performance due to discrepancies in learning objectives (recommendation vs. noise reconstruction) and the respective learning spaces (non-stationary vs. stationary). To overcome this, we propose a novel framework called DimeRec (Di ffusion with multi-interest enhanced Rec ommender). DimeRec synergistically combines a guidance extraction module (GEM) and a generative diffusion aggregation module (DAM). The GEM extracts crucial stationary guidance signals from the user's non-stationary interaction history, while the DAM employs a generative diffusion process conditioned on GEM's outputs to reconstruct and generate consistent recommendations. Our numerical experiments demonstrate that DimeRec significantly outperforms established baseline methods across three publicly available datasets. Furthermore, we have successfully deployed DimeRec on a large-scale short video recommendation platform, serving hundreds of millions of users. Live A/B testing confirms that our method improves both users' time spent and result diversification.

Scale-calibrated ranking systems are ubiquitous in real-world applications nowadays, which pursue accurate ranking quality and calibrated probabilistic predictions simultaneously.For instance, in the advertising ranking system, the predicted click-through rate (CTR) is utilized for ranking and required to be calibrated for the downstream cost-per-click ads bidding.Recently, multi-objective based methods have been wildly adopted as a standard approach for Calibrated Ranking, which incorporates the combination of two loss functions: a pointwise loss that focuses on calibrated absolute values and a ranking loss that emphasizes relative orderings.However, when applied to industrial online applications, existing multi-objective CR approaches still suffer from two crucial limitations.First, previous methods need to aggregate the full candidate list within a single mini-batch to compute the ranking loss.Such aggregation strategy violates extensive data shuffling which has long been proven beneficial for preventing overfitting, and thus degrades the training effectiveness.Second, existing multi-objective methods apply the two inherently conflicting loss functions on a single probabilistic prediction, which results in a sub-optimal trade-off between calibration and ranking.To tackle the two limitations, we propose a Self-Boosted framework for Calibrated Ranking (SBCR).In SBCR, the predicted ranking scores by the online deployed model are dumped into context features.With these additional context features, each single item can perceive the overall distribution of scores in the whole ranking list, so that the ranking loss can be constructed without the need for sample aggregation.As the deployed model is a few versions older than the training model, the dumped predictions reveal what was failed to learn and keep boosting the model to correct previously mis-predicted items.Moreover, a calibration module is introduced to decouple the point loss and ranking loss.The two losses are applied before and after the calibration module separately, which

Modeling user's historical feedback is essential for Click-Through Rate Prediction in personalized search and recommendation. Existing methods usually only model users' positive feedback information such as click sequences which neglects the context information of the feedback. In this paper, we propose a new perspective for context-aware users' behavior modeling by including the whole page-wisely exposed products and the corresponding feedback as contextualized page-wise feedback sequence. The intra-page context information and inter-page interest evolution can be captured to learn more specific user preference. We design a novel neural ranking model RACP(Recurrent Attention over Contextualized Page sequence), which utilizes page-context aware attention to model the intra-page context. A recurrent attention process is used to model the cross-page interest convergence evolution as denoising the interest in the previous pages. Experiments on public and real-world industrial datasets verify our model's effectiveness.

Multi-scenario Learning to Rank is essential for Recommender Systems, Search Engines and Online Advertising in e-commerce portals where the ranking models are usually applied in many scenarios. However, existing works mainly focus on learning the ranking model for a single scenario, and pay less attention to learning ranking models for multiple scenarios. We identify two practical challenges in industrial multi-scenario ranking systems: (1) The Feedback Loop problem that the model is always trained on the items chosen by the ranker itself. (2) Insufficient training data for small and new scenarios. To address the above issues, we present ZEUS, a novel framework that learns a Zoo of ranking modEls for mUltiple Scenarios based on pre-training on users' spontaneous behaviors (e.g. queries which are directly searched in the search box and not recommended by the ranking system). ZEUS decomposes the training process into two stages: self-supervised learning based pre-training and fine-tuning. Firstly, ZEUS performs self-supervised learning on users' spontaneous behaviors and generates a pre-trained model. Secondly, ZEUS fine-tunes the pre-trained model on users' implicit feedback in multiple scenarios. Extensive experiments on Alibaba's production dataset demonstrate the effectiveness of ZEUS, which significantly outperforms state-of-the-art methods. ZEUS averagely achieves 6.0%, 9.7%, 11.7% improvement in CTR, CVR and GMV respectively than state-of-the-art method.

Conversion Rate (CVR) prediction in modern industrial e-commerce platforms is becoming increasingly important, which directly contributes to the final revenue. In order to address the well-known sample selection bias (SSB) and data sparsity (DS) issues encountered during CVR modeling, the abundant labeled macro behaviors (i.e., user's interactions with items) are used. Nonetheless, we observe that several purchase-related micro behaviors (i.e., user's interactions with specific components on the item detail page) can supplement fine-grained cues for CVR prediction. Motivated by this observation, we propose a novel CVR prediction method by Hierarchically Modeling both Micro and Macro behaviors (HM3). Specifically, we first construct a complete user sequential behavior graph to hierarchically represent micro behaviors and macro behaviors as one-hop and two-hop post-click nodes. Then, we embody HM3 as a multi-head deep neural network, which predicts six probability variables corresponding to explicit sub-paths in the graph. They are further combined into the prediction targets of four auxiliary tasks as well as the final CVR according to the conditional probability rule defined on the graph. By employing multi-task learning and leveraging the abundant supervisory labels from micro and macro behaviors, HM3 can be trained end-to-end and address the SSB and DS issues. Extensive experiments on both offline and online settings demonstrate the superiority of the proposed HM3 over representative state-of-the-art methods.

Purchase-related micro-behaviors, e.g., favorite, add to cart, read reviews, etc., provide implicit feedback of users' decision-making process. Such informative feedback can lead to fine-grained post-click conversion rate (CVR) modeling of the buying process. However, most existing works on CVR estimation either neglect these informative feedback, or model them as a sequential pattern with Recurrent Neural Networks. We argue such modeling could be inappropriate since different orders of micro-behaviors may represent similar user buying intention, and micro-behaviors often correlate with each other.

Recommender system, as an essential part of modern e-commerce, consists of two fundamental modules, namely Click-Through Rate (CTR) and Conversion Rate (CVR) prediction. While CVR has a direct impact on the purchasing volume, its prediction is well-known challenging due to the Sample Selection Bias (SSB) and Data Sparsity (DS) issues. Although existing methods, typically built on the user sequential behavior path "impression->click->purchase", is effective for dealing with SSB issue, they still struggle to address the DS issue due to rare purchase training samples. Observing that users always take several purchase-related actions after clicking, we propose a novel idea of post-click behavior decomposition. Specifically, disjoint purchase-related Deterministic Action (DAction) and Other Action (OAction) are inserted between click and purchase in parallel, forming a novel user sequential behavior graph "impression->click->D(O)Action->purchase". Defining model on this graph enables to leverage all the impression samples over the entire space and extra abundant supervised signals from D(O)Action, which will effectively address the SSB and DS issues together. To this end, we devise a novel deep recommendation model named Elaborated Entire Space Supervised Multi-task Model (ESM2). According to the conditional probability rule defined on the graph, it employs multi-task learning to predict some decomposed sub-targets in parallel and compose them sequentially to formulate the final CVR. Extensive experiments on both offline and online environments demonstrate the superiority of ESM2 over state-of-the-art models. The source code and dataset will be released.

Post-click conversion rate (CVR) estimation is a critical task in e-commerce recommender systems. This task is deemed quite challenging under industrial setting with two major issues: 1) selection bias caused by user self-selection, and 2) data sparsity due to the rare click events. A successful conversion typically has the following sequential events: ”exposure → click → conversion”. Conventional CVR estimators are trained in the click space, but inference is done in the entire exposure space. They fail to account for the causes of the missing data and treat them as missing at random. Hence, their estimations are highly likely to deviate from the real values by large. In addition, the data sparsity issue can also handicap many industrial CVR estimators which usually have large parameter spaces.