Short-video recommender systems typically optimize ranking models using dense user behavioral signals, such as clicks and watch time. However, these signals are only indirect proxies of user satisfaction and often suffer from noise and bias. Recently, explicit satisfaction feedback collected through questionnaires has emerged as a high-quality direct alignment supervision, but is extremely sparse and easily overwhelmed by abundant behavioral data, making it difficult to incorporate into online recommendation models. To address these challenges, we propose a novel framework which is towards End-to-End Alignment of user Satisfaction via Questionnaire, named EASQ, to enable real-time alignment of ranking models with true user satisfaction. Specifically, we first construct an independent parameter pathway for sparse questionnaire signals by combining a multi-task architecture and a lightweight LoRA module. The multi-task design separates sparse satisfaction supervision from dense behavioral signals, preventing the former from being overwhelmed. The LoRA module pre-inject these preferences in a parameter-isolated manner, ensuring stability in the backbone while optimizing user satisfaction. Furthermore, we employ a DPO-based optimization objective tailored for online learning, which aligns the main model outputs with sparse satisfaction signals in real time. This design enables end-to-end online learning, allowing the model to continuously adapt to new questionnaire feedback while maintaining the stability and effectiveness of the backbone. Extensive offline experiments and large-scale online A/B tests demonstrate that EASQ consistently improves user satisfaction metrics across multiple scenarios. EASQ has been successfully deployed in a production short-video recommendation system, delivering significant and stable business gains.

We present PushGen, an automated framework for generating high-quality push notifications comparable to human-crafted content. With the rise of generative models, there is growing interest in leveraging LLMs for push content generation. Although LLMs make content generation straightforward and cost-effective, maintaining stylistic control and reliable quality assessment remains challenging, as both directly impact user engagement. To address these issues, PushGen combines two key components: (1) a controllable category prompt technique to guide LLM outputs toward desired styles, and (2) a reward model that ranks and selects generated candidates. Extensive offline and online experiments demonstrate its effectiveness, which has been deployed in large-scale industrial applications, serving hundreds of millions of users daily.

Post-click conversion rate (CVR) estimation is a vital task in many recommender systems of revenue businesses, e.g., e-commerce and advertising. In a perspective of sample, a typical CVR positive sample usually goes through a funnel of exposure?click?conversion. For lack of post-event labels for un-clicked samples, CVR learning task commonly only utilizes clicked samples, rather than all exposed samples as for click-through rate (CTR) learning task. However, during online inference, CVR and CTR are estimated on the same assumed exposure space, which leads to a inconsistency of sample space between training and inference, i.e., sample selection bias (SSB). To alleviate SSB, previous wisdom proposes to design novel auxiliary tasks to enable the CVR learning on un-click training samples, such as CTCVR and counterfactual CVR, etc. Although alleviating SSB to some extent, none of them pay attention to the discrimination between ambiguous negative samples (un-clicked) and factual negative samples (clicked but un-converted) during modelling, which makes CVR model lacks robustness. To full this gap, we propose a novel ChorusCVR model to realize debiased CVR learning in entire-space. We propose a Negative sample Discrimination Module (NDM), which aims to provide robust soft labels with the ability to discriminate factual negative samples (clicked but un-converted) from ambiguous negative samples (un-clicked). Moreover, we propose a Soft Alignment Module (SAM) to supervise CVR learning with several alignment objectives using generated soft labels. Extensive offline experiments and online A/B testing at Kuaishou's e-commerce live service validates our ChorusCVR.

Local life service is a vital scenario in Kuaishou App, in which we recommend videos with stores' location information. Thus, recommendation in our scenario is challenging because we should take into account user's interest and real-time location at the same time. In complex scenarios, end-to-end generative recommendation has emerged as a new paradigm, such as OneRec in the short video scenario, OneSug in the search scenario, and EGA in the advertising scenario. However, in local life service, an end-to-end generative recommendation model has not yet been developed as there are some key challenges to be solved. The first challenge is how to make full use of geographic information. The second challenge is how to balance multiple objectives, including user interests, the distance between user and stores, and some other business objectives. To address the challenges, we propose OneLoc. Specifically, we leverage geographic information from different perspectives: (1) geo-aware semantic ID incorporates both video and geographic information for tokenization, (2) geo-aware attention injects video location similarity and user's real-time location in the encoder, and (3) neighbor-aware prompt captures rich context information surrounding users for generation. To balance multiple objectives, we use reinforcement learning and propose a geographic reward and a GMV reward. With the above design, OneLoc achieves outstanding offline and online performance. In fact, OneLoc has been deployed in local life service of Kuaishou App and achieved 21.016% and 17.891% improvements in terms of gross merchandise value (GMV) and orders numbers.

Search-based recommendation is one of the most critical application scenarios in e-commerce platforms. Users' complex search contexts—such as spatiotemporal factors, historical interactions, and current query's information—constitute an essential part of their decision-making, reflecting implicit preferences that complement explicit query terms. Modeling such rich contextual signals and their intricate associations with candidate items remains a key challenge. Although numerous efforts have been devoted to building more effective search methods, existing approaches still show limitations in integrating contextual information, which hinders their ability to fully capture user intent. To address these challenges, we propose a context-aware reasoning-enhanced generative search framework for better understanding the complicated context. Specifically, the framework first unifies heterogeneous user and item contexts into textual representations or text-based semantic identifiers and aligns them. To overcome the lack of explicit reasoning trajectories, we introduce a self-evolving post-training paradigm that iteratively combines supervised fine-tuning and reinforcement learning to progressively enhance the model's reasoning capability. In addition, we identify potential biases in existing RL algorithms when applied to search scenarios and present a debiased variant of GRPO to improve ranking performance. Extensive experiments on search log data collected from a real-world e-commerce platform demonstrate that our approach achieves superior performance compared with strong baselines, validating its effectiveness for search-based recommendation.

Industrial-scale recommender systems rely on a cascade pipeline in which the retrieval stage must return a high-recall candidate set from billions of items under tight latency. Existing solutions either (i) suffer from limited expressiveness in capturing fine-grained user-item interactions, as seen in decoupled dual-tower architectures that rely on separate encoders, or generative models that lack precise target-aware matching capabilities, or (ii) build structured indices (tree, graph, quantization) whose item-centric topologies struggle to incorporate dynamic user preferences and incur prohibitive construction and maintenance costs.

To provide promising recommendation results, there exist three major stages in the industrial RecSys chain to support our service: (1) The first Retrieval model aims at searching hundreds of item candidates. (2) Next, the Ranking model estimates the multiple aspect probabilities Pxtrs for each retrieved item. (3) At last, the Ensemble Sort stage merges those Pxtrs into one comparable score, and then selects the best dozen items with the highest scores to recommend them. To our knowledge, the wide-accepted industry ensemble sort approach still relies on manual formula-based adjustment, i.e., assigning manual weights for Pxtrs to control its influence on fusion score. Under this framework, the RecSys severely relies on expert knowledge to determine satisfactory weight for each Pxtr, which blocks RecSys's further advancements.

Large-scale industrial recommendation systems typically employ a two-stage paradigm of retrieval and ranking to handle huge amounts of information. Recent research focuses on improving the performance of retrieval model. A promising way is to introduce extensive information about users and items. On one hand, lifelong sequential behavior is valuable. Existing lifelong behavior modeling methods in ranking stage focus on the interaction of lifelong behavior and candidate items from retrieval stage. In retrieval stage, it is difficult to utilize lifelong behavior because of a large corpus of candidate items. On the other hand, existing retrieval methods mostly relay on interaction information, potentially disregarding valuable multi-modal information. To solve these problems, we represent the pioneering exploration of leveraging multi-modal information and lifelong sequence model within the advanced tree-based retrieval model. We propose Multi-modal Indexing and Searching with lifelong Sequence (MISS), which contains a multi-modal index tree and a multi-modal lifelong sequence modeling module. Specifically, for better index structure, we propose multi-modal index tree, which is built using the multi-modal embedding to precisely represent item similarity. To precisely capture diverse user interests in user lifelong sequence, we propose collaborative general search unit (Co-GSU) and multi-modal general search unit (MM-GSU) for multi-perspective interests searching.

Modern industrial recommendation systems encounter a core chal- lenge of multi-stage optimization misalignment: a significant semantic gap exists between the multi-objective optimization paradigm (such as jointly optimizing click-through rate, watch duration, and conversion rate) widely used in the ranking phase and the single-objective modeling in the retrieve phase. Although the main-stream industry solution achieves multi-objective coverage throughparallel multi-path single-objective retrieve, this approach leads to linear growth of training and serving resources with the number of objectives and has inherent limitations in handling loosely coupled objectives. This paper proposes the MPFormer, a dynamic multi-task Transformer framework, which systematically addresses the aforementioned issues through three innovative mechanisms. First, an objective-conditioned transformer that jointly encodes user behavior sequences and multi-task semantics through learnable attention modulation; second, personalized target weights are introduced to achieve dynamic adjustment of retrieve results; finally, user personalization information is incorporated into token representations and the Transformer structure to further enhance the model's representation ability. This framework has been successfully integrated into Kuaishou's short video recommendation system, stably serving over 400 million daily active users. It significantly improves user daily engagement and system operational efficiency. Practical deployment verification shows that, compared with traditional solutions, it effectively optimizes the multi-objective retrieve iteration paradigm while maintaining service response speed, providing a scalable multi-objective solution for industrial recommendation systems.

In this paper, we present the practical problems and the lessons learned at short-video services from Kuaishou. In industry, a widely-used multi-task framework is the Mixture-of-Experts (MoE) paradigm, which always introduces some shared and specific experts for each task and then uses gate networks to measure related experts' contributions. Although the MoE achieves remarkable improvements, we still observe three anomalies that seriously affect model performances in our iteration: (1) Expert Collapse: We found that experts' output distributions are significantly different, and some experts have over 90% zero activations with ReLU, making it hard for gate networks to assign fair weights to balance experts. (2) Expert Degradation: Ideally, the shared-expert aims to provide predictive information for all tasks simultaneously. Nevertheless, we find that some shared-experts are occupied by only one task, which indicates that shared-experts lost their ability but degenerated into some specific-experts. (3) Expert Underfitting: In our services, we have dozens of behavior tasks that need to be predicted, but we find that some data-sparse prediction tasks tend to ignore their specific-experts and assign large weights to shared-experts. The reason might be that the shared-experts can perceive more gradient updates and knowledge from dense tasks, while specific-experts easily fall into underfitting due to their sparse behaviors.

The popularity of short video applications has brought new opportunities and challenges to video recommendation.In addition to the traditional ranking-based pipeline, industrial solutions usually introduce additional distribution management components to guarantee a diverse and content-rich user experience.However, existing solutions are either non-personalized or fail to generalize well to the ever-changing user preferences.Inspired by the success of uplift modeling in online marketing, we attempt to implement uplift modeling in the video recommendation scenario to mitigate the problems.However, we face two main challenges when migrating the technique: 1) the complex-response causal relation in distribution management problem, and 2) the modeling of long-term and real-time user preferences.To address these challenges, we correspond each treatment to a specific adjustment of the distribution over video types, then propose a Coarse-to-fine Dynamic Uplift Modeling (CDUM) framework for real-time video recommendation scenarios.Specifically, CDUM consists of two modules, a coarse-grained module that utilizes the offline features of users to model their long-term preferences, and a fine-grained * Both authors contributed equally to this research.

Auto-bidding, with its strong capability to optimize bidding decisions within dynamic and competitive online environments, has become a pivotal strategy for advertising platforms. Existing approaches typically employ rule-based strategies or Reinforcement Learning (RL) techniques. However, rule-based strategies lack the flexibility to adapt to time-varying market conditions, and RL-based methods struggle to capture essential historical dependencies and observations within Markov Decision Process (MDP) frameworks. Furthermore, these approaches often face challenges in ensuring strategy adaptability across diverse advertising objectives. Additionally, as offline training methods are increasingly adopted to facilitate the deployment and maintenance of stable online strategies, the issues of documented behavioral patterns and behavioral collapse resulting from training on fixed offline datasets become increasingly significant. To address these limitations, this paper introduces a novel offline Generative Auto-bidding framework with Value-Guided Explorations (GAVE). GAVE accommodates various advertising objectives through a score-based Return-To-Go (RTG) module. Moreover, GAVE integrates an action exploration mechanism with an RTG-based evaluation method to explore novel actions while ensuring stability-preserving updates. A learnable value function is also designed to guide the direction of action exploration and mitigate Out-of-Distribution (OOD) problems. Experimental results on two offline datasets and real-world deployments demonstrate that GAVE outperforms state-of-the-art baselines in both offline evaluations and online A/B tests. By applying the core methods of this framework, we proudly secured first place in the NeurIPS 2024 competition, 'AIGB Track: Learning Auto-Bidding Agents with Generative Models'.

People's daily lives involve numerous periodic behaviors, such as eating and traveling. Local-life platforms cater to these recurring needs by providing essential services tied to daily routines. Therefore, users' periodic intentions are reflected in their interactions with the platforms. There are two main challenges in modeling users' periodic behaviors in the local-life service recommendation systems: 1) the diverse demands of users exhibit varying periodicities, which are difficult to distinguish as they are mixed in the behavior sequences; 2) the periodic behaviors of users are subject to dynamic changes due to factors such as holidays and promotional events. Existing methods struggle to distinguish the periodicities of diverse demands and overlook the importance of dynamically capturing changes in users' periodic behaviors. To this end, we employ a Frequency-Aware Multi-View Interest Modeling framework (FIM). Specifically, we propose a multi-view search strategy that decomposes users' demands from different perspectives to separate their various periodic intentions. This allows the model to comprehensively extract their periodic features than category-searched-only methods. Moreover, we propose a frequency-domain perception and evolution module. This module uses the Fourier Transform to convert users' temporal behaviors into the frequency domain, enabling the model to dynamically perceive their periodic features. Extensive offline experiments demonstrate that FIM achieves significant improvements on public and industrial datasets, showing its capability to effectively model users' periodic intentions. Furthermore, the model has been deployed on the Kuaishou local-life service platform. Through online A/B experiments, the transaction volume has been significantly improved.

Large Language Models (LLMs) have garnered significant attention in Recommendation Systems (RS) due to their extensive world knowledge and robust reasoning capabilities. However, a critical challenge lies in enabling LLMs to effectively comprehend and extract insights from massive user behaviors. Current approaches that directly leverage LLMs for user interest learning face limitations in handling long sequential behaviors, effectively extracting interest, and applying interest in practical scenarios. To address these issues, we propose a Hierarchical Tree Search-based User Lifelong Behavior Modeling framework (HiT-LBM). HiT-LBM integrates Chunked User Behavior Extraction (CUBE) and Hierarchical Tree Search for Interest (HTS) to capture diverse interests and interest evolution of user. CUBE divides user lifelong behaviors into multiple chunks and learns the interest and interest evolution within each chunk in a cascading manner. HTS generates candidate interests through hierarchical expansion and searches for the optimal interest with process rating model to ensure information gain for each behavior chunk. Additionally, we design Temporal-Ware Interest Fusion (TIF) to integrate interests from multiple behavior chunks, constructing a comprehensive representation of user lifelong interests. The representation can be embedded into any recommendation model to enhance performance. Extensive experiments demonstrate the effectiveness of our approach, showing that it surpasses state-of-the-art methods. We also deploy HiT-LBM on Kuaishou's online advertising platform, showing 3.5% increase of revenue. We release the implementation code https://github.com/xiayu-cell/HiT-LBM.

Reranking models solve the final recommendation lists that best fulfill users' demands.While existing solutions focus on finding parametric models that approximate optimal policies, recent approaches find that it is better to generate multiple lists to compete for a "pass" ticket from an evaluator, where the evaluator serves as the supervisor who accurately estimates the performance of the candidate lists.In this work, we show that we can achieve a more efficient and effective list proposal with a multi-generator framework and provide empirical evidence on two public datasets and online A/B tests.More importantly, we verify that the effectiveness of a generator is closely related to how much it complements the views of other generators with sufficiently different rerankings, which derives the metric of list comprehensiveness.With this intuition, we design an automatic complementary generator-finding framework that learns a policy that simultaneously aligns the users' preferences and maximizes the list comprehensiveness metric.The experimental results indicate that the proposed framework can further improve the multi-generator reranking performance.

The field of Reinforcement Learning (RL) has garnered increasing attention for its ability of optimizing user retention in recommender systems. A primary obstacle in this optimization process is the environment non-stationarity stemming from the continual and complex evolution of user behavior patterns over time, such as variations in interaction rates and retention propensities. These changes pose significant challenges to existing RL algorithms for recommendations, leading to issues with dynamics and reward distribution shifts. This paper introduces a novel approach called Adaptive User Retention Optimization (AURO) to address this challenge. To navigate the recommendation policy in non-stationary environments, AURO introduces an state abstraction module in the policy network. The module is trained with a new value-based loss function, aligning its output with the estimated performance of the current policy. As the policy performance of RL is sensitive to environment drifts, the loss function enables the state abstraction to be reflective of environment changes and notify the recommendation policy to adapt accordingly. Additionally, the non-stationarity of the environment introduces the problem of implicit cold start, where the recommendation policy continuously interacts with users displaying novel behavior patterns. AURO encourages exploration guarded by performance-based rejection sampling to maintain a stable recommendation quality in the cost-sensitive online environment. Extensive empirical analysis are conducted in a user retention simulator, the MovieLens dataset, and a live short-video recommendation platform, demonstrating AURO's superior performance against all evaluated baseline algorithms. Code is available at https://github.com/AIDefender/AURO

Rich user behavior data has been proven to be of great value for recommendation systems. Modeling lifelong user behavior data in the retrieval stage to explore user long-term preference and obtain comprehensive retrieval results is crucial. Existing lifelong modeling methods cannot applied to the retrieval stage because they extract target-relevant items through the coupling between the user and the target item. Moreover, the current retrieval methods fail to precisely capture user interests when the length of the user behavior sequence increases further. That leads to a gap in the ability of retrieval models to model lifelong user behavior data. In this paper, we propose the concept of missing interest, leveraging the idea of complementarity, which serves as a supplement to short-term interest based on lifelong behavior data in the retrieval stage. Specifically, we design a missing interest operator and deploy it in Kafka data stream, without incurring latency or storage costs. This operator derives categories and authors of items that the user was previously interested in but has recently missed, and uses these as triggers to output missing features to the downstream retrieval model. Our retrieval model is a complete dual-tower structure that combines short-term and missing interests on the user side to provide a comprehensive depiction of lifelong behaviors. Since 2023, the presented solution has been deployed in Kuaishou, one of the most popular short-video streaming platforms in China with hundreds of millions of active users.

Achieving fairness among different individuals or groups is an essential task for industrial recommender systems. Due to the group's personalized selection tendencies and the non-uniform population distributions, existing industrial recommenders tend to make unfair predictions towards the preferences of minority groups. To alleviate this unfairness, we propose a model-agnostic self-adaptive fairness constraint framework (SaFair) based on the posterior preferences of different groups. We construct group-level and individual-level fairness constraints. The former measures consistency between group-level posterior preferences and predicted interests, and the latter relies on the degree of consistency in interests between a user and their associated group to perform self-adaptive constraints. In particular, to balance effectiveness and fairness, we utilize uncertainty estimation to adjust the intensity of constraints according to the model's learning status called self-adaptive constraints. Extensive offline experiments and online A/B Testing are conducted and the results validate the superiority of our proposed method over the baselines. SaFair has been successfully deployed in Kuaishou, one of China's most popular short-video streaming platforms with hundreds of millions of active users.

In large-scale recommendation systems, modeling long-term user interests is progressively gaining attention among researchers and practitioners. Existing work, such as SIM and TWIN, typically employs a two-stage approach to model long-term user behavior sequences for efficiency concerns. The first stage rapidly retrieves a subset of sequences related to the target item from a long sequence using a search-based mechanism namely the General Search Unit (GSU), while the second stage calculates the interest scores using the Exact Search Unit (ESU) on the retrieved results. Given the extensive length of user behavior sequences spanning the entire life cycle, potentially reaching up to 10^6 in scale, there is currently no effective solution for fully modeling such expansive user interests. To overcome this issue, we introduced TWIN-V2, an enhancement of TWIN, where a divide-and-conquer approach is applied to compress life-cycle behaviors and uncover more accurate and diverse user interests. Specifically, a hierarchical clustering method groups items with similar characteristics in life-cycle behaviors into a single cluster during the offline phase. By limiting the size of clusters, we can compress behavior sequences well beyond the magnitude of 10^5 to a length manageable for online inference in GSU retrieval. Cluster-aware target attention extracts comprehensive and multi-faceted long-term interests of users, thereby making the final recommendation results more accurate and diverse. Extensive offline experiments on a multi-billion-scale industrial dataset and online A/B tests have demonstrated the effectiveness of TWIN-V2. Under an efficient deployment framework, TWIN-V2 has been successfully deployed to the primary traffic that serves hundreds of millions of daily active users at Kuaishou.

In recommender systems, reinforcement learning solutions have shown promising results in optimizing the interaction sequence between users and the system over the long-term performance. For practical reasons, the policy's actions are typically designed as recommending a list of items to handle users' frequent and continuous browsing requests more efficiently. In this list-wise recommendation scenario, the user state is updated upon every request in the corresponding MDP formulation. However, this request-level formulation is essentially inconsistent with the user's item-level behavior. In this study, we demonstrate that an item-level optimization approach can better utilize item characteristics and optimize the policy's performance even under the request-level MDP. We support this claim by comparing the performance of standard request-level methods with the proposed item-level actor-critic framework in both simulation and online experiments. Furthermore, we show that a reward-based future decomposition strategy can better express the item-wise future impact and improve the recommendation accuracy in the long term. To achieve a more thorough understanding of the decomposition strategy, we propose a model-based re-weighting framework with adversarial learning that further boost the performance and investigate its correlation with the reward-based strategy.

Short video has witnessed rapid growth in the past few years in multimedia platforms. To ensure the freshness of the videos, platforms receive a large number of user-uploaded videos every day, making collaborative filtering-based recommender methods suffer from the item cold-start problem (e.g., the new-coming videos are difficult to compete with existing videos). Consequently, increasing efforts tackle the cold-start issue from the content perspective, focusing on modeling the multi-modal preferences of users, a fair way to compete with new-coming and existing videos. However, recent studies ignore the existing gap between multi-modal embedding extraction and user interest modeling as well as the discrepant intensities of user preferences for different modalities. In this paper, we propose M3CSR, a multi-modal modeling framework for cold-start short video recommendation. Specifically, we preprocess content-oriented multi-modal features for items and obtain trainable category IDs by performing clustering. In each modality, we combine modality-specific cluster ID embedding and the mapped original modality feature as modality-specific representation of the item to address the gap. Meanwhile, M3CSR measures the user modality-specific intensity based on the correlation between modality-specific interest and behavioral interest and employs pairwise loss to further decouple user multi-modal interests. Extensive experiments on four real-world datasets demonstrate the superiority of our proposed model. The framework has been deployed on a billion-user scale short video application and has shown improvements in various commercial metrics within cold-start scenarios.

Modern recommender systems are built upon computation-intensive infrastructure, and it is challenging to perform real-time computation for each request, especially in peak periods, due to the limited computational resources. Recommending by user-wise result caches is widely used when the system cannot afford a real-time recommendation. However, it is challenging to allocate real-time and cached recommendations to maximize the users’ overall engagement. This paper shows two key challenges to cache allocation, i.e., the value-strategy dependency and the streaming allocation. Then, we propose a reinforcement prediction-allocation framework (RPAF) to address these issues. RPAF is a reinforcement-learning-based two-stage framework containing prediction and allocation stages. The prediction stage estimates the values of the cache choices considering the value-strategy dependency, and the allocation stage determines the cache choices for each individual request while satisfying the global budget constraint. We show that the challenge of training RPAF includes globality and the strictness of budget constraints, and a relaxed local allocator (RLA) is proposed to address this issue. Moreover, a PoolRank algorithm is used in the allocation stage to deal with the streaming allocation problem. Experiments show that RPAF significantly improves users’ engagement under computational budget constraints.

Recommender systems filter out information that meets user interests. However, users may be tired of the recommendations that are too similar to the content they have been exposed to in a short historical period, which is the so-called user fatigue. Despite the significance for a better user experience, user fatigue is seldom explored by existing recommenders. In fact, there are three main challenges to be addressed for modeling user fatigue, including what features support it, how it influences user interests, and how its explicit signals are obtained. In this paper, we propose to model user Fatigue in interest learning for sequential Recommendations (FRec). To address the first challenge, based on a multi-interest framework, we connect the target item with historical items and construct an interest-aware similarity matrix as features to support fatigue modeling. Regarding the second challenge, built upon feature cross, we propose a fatigue-enhanced multi-interest fusion to capture long-term interest. In addition, we develop a fatigue-gated recurrent unit for short-term interest learning, with temporal fatigue representations as important inputs for constructing update and reset gates. For the last challenge, we propose a novel sequence augmentation to obtain explicit fatigue signals for contrastive learning. We conduct extensive experiments on real-world datasets, including two public datasets and one large-scale industrial dataset. Experimental results show that FRec can improve AUC and GAUC up to 0.026 and 0.019 compared with state-of-the-art models, respectively. Moreover, large-scale online experiments demonstrate the effectiveness of FRec for fatigue reduction. Our codes are released at https://github.com/tsinghua-fib-lab/SIGIR24-FRec.

In Recommender System (RS) applications, reinforcement learning (RL) has recently emerged as a powerful tool, primarily due to its proficiency in optimizing long-term rewards. Nevertheless, it suffers from instability in the learning process, stemming from the intricate interactions among bootstrapping, off-policy training, and function approximation. Moreover, in multi-reward recommendation scenarios, designing a proper reward setting that reconciles the inner dynamics of various tasks is quite intricate. To this end, we propose a novel decision transformer-based recommendation model, DT4IER, to not only elevate the effectiveness of recommendations but also to achieve a harmonious balance between immediate user engagement and long-term retention. The DT4IER applies an innovative multi-reward design that adeptly balances short and long-term rewards with user-specific attributes, which serve to enhance the contextual richness of the reward sequence, ensuring a more informed and personalized recommendation process. To enhance its predictive capabilities, DT4IER incorporates a high-dimensional encoder to identify and leverage the intricate interrelations across diverse tasks. Furthermore, we integrate a contrastive learning approach within the action embedding predictions, significantly boosting the model's overall performance. Experiments on three real-world datasets demonstrate the effectiveness of DT4IER against state-of-the-art baselines in terms of both immediate user engagement and long-term retention. The source code is accessible online to facilitate replication.

Multi-domain recommendation and multi-task recommendation have demonstrated their effectiveness in leveraging common information from different domains and objectives for comprehensive user modeling. Nonetheless, the practical recommendation usually faces multiple domains and tasks simultaneously, which cannot be well-addressed by current methods. To this end, we introduce M3oE, an adaptive Multi-domain Multi-task Mixture-of-Experts recommendation framework. M3oE integrates multi-domain information, maps knowledge across domains and tasks, and optimizes multiple objectives. We leverage three mixture-of-experts modules to learn common, domain-aspect, and task-aspect user preferences respectively to address the complex dependencies among multiple domains and tasks in a disentangled manner. Additionally, we design a two-level fusion mechanism for precise control over feature extraction and fusion across diverse domains and tasks. The framework's adaptability is further enhanced by applying AutoML technique, which allows dynamic structure optimization. To the best of the authors' knowledge, our M3oE is the first effort to solve multi-domain multi-task recommendation self-adaptively. Extensive experiments on two benchmark datasets against diverse baselines demonstrate M3oE's superior performance. The implementation code is available to ensure reproducibility.

In modern recommender systems, sequential recommendation leverages chronological user behaviors to make effective next-item suggestions, which suffers from data sparsity issues, especially for new users. One promising line of work is the cross-domain recommendation, which trains models with data across multiple domains to improve the performance in data-scarce domains. Recent proposed cross-domain sequential recommendation models such as PiNet and DASL have a common drawback relying heavily on overlapped users in different domains, which limits their usage in practical recommender systems. In this paper, we propose a M ixed A ttention N etwork (MAN) with local and global attention modules to extract the domain-specific and cross-domain information. Firstly, we propose a local/global encoding layer to capture the domain-specific/cross-domain sequential pattern. Then we propose a mixed attention layer with item similarity attention, sequence-fusion attention, and group-prototype attention to capture the local/global item similarity, fuse the local/global item sequence, and extract the user groups across different domains, respectively. Finally, we propose a local/global prediction layer to further evolve and combine the domain-specific and cross-domain interests. Experimental results on two real-world datasets (each with two domains) demonstrate the superiority of our proposed model. Further study also illustrates that our proposed method and components are model-agnostic and effective, respectively. The code and data are available at https://github.com/Guanyu-Lin/MAN.

Modern personalized recommendation services often rely on user feedback, either explicit or implicit, to improve the quality of services. Explicit feedback refers to behaviors like ratings, while implicit feedback refers to behaviors like user clicks. However, in the scenario of full-screen video viewing experiences like Tiktok and Reels, the click action is absent, resulting in unclear feedback from users, hence introducing noises in modeling training. Existing approaches on de-noising recommendation mainly focus on positive instances while ignoring the noise in a large amount of sampled negative feedback. In this paper, we propose a meta-learning method to annotate the unlabeled data from loss and gradient perspectives, which considers the noises in both positive and negative instances. Specifically, we first propose anInverse Dual Loss (IDL) to boost the true label learning and prevent the false label learning. Then we further propose anInverse Gradient (IG) method to explore the correct updating gradient and adjust the updating based on meta-learning. Finally, we conduct extensive experiments on both benchmark and industrial datasets where our proposed method can significantly improve AUC by 9.25% against state-of-the-art methods. Further analysis verifies the proposed inverse learning framework is model-agnostic and can improve a variety of recommendation backbones. The source code, along with the best hyper-parameter settings, is available at this link: https://github.com/Guanyu-Lin/InverseLearning.

Cascade ranking is widely used for large-scale top-k selection problems in online advertising and recommendation systems, and learning-to-rank is an important way to optimize the models in cascade ranking. Previous works on learning-to-rank usually focus on letting the model learn the complete order or top-k order, and adopt the corresponding rank metrics (e.g. OPA and NDCG@k) as optimization targets. However, these targets can not adapt to various cascade ranking scenarios with varying data complexities and model capabilities; and the existing metric-driven methods such as the Lambda framework can only optimize a rough upper bound of limited metrics, potentially resulting in sub-optimal and performance misalignment. To address these issues, we propose a novel perspective on optimizing cascade ranking systems by highlighting the adaptability of optimization targets to data complexities and model capabilities. Concretely, we employ multi-task learning to adaptively combine the optimization of relaxed and full targets, which refers to metrics Recall@m@k and OPA respectively. We also introduce permutation matrix to represent the rank metrics and employ differentiable sorting techniques to relax hard permutation matrix with controllable approximate error bound. This enables us to optimize both the relaxed and full targets directly and more appropriately. We named this method as Adaptive Neural Ranking Framework (abbreviated as ARF). Furthermore, we give a specific practice under ARF. We use the NeuralSort to obtain the relaxed permutation matrix and draw on the variant of the uncertainty weight method in multi-task learning to optimize the proposed losses jointly. Experiments on a total of 4 public and industrial benchmarks show the effectiveness and generalization of our method, and online experiment shows that our method has significant application value.

Recommender systems have emerged as an indispensable mean to meet personalized interests of users and alleviate information overload. Despite the great success, accuracy-oriented recommendation models are creating information cocoons, i.e., it is becoming increasingly difficult for users to see other items they might be interested in. Although recent studies start paying attention to enhancing recommendation diversity, models based on point embedding fail to describe the range of user preferences and item features well, which is essential for diversified matching. To this end, we propose LCD-UC , a novel List-Check-Decide framework with UnCertainty masking based on box embedding to improve recommendation diversity with recommendation accuracy maintained. Specifically, LCD-UC creates hypercubes to represent users and items using box embedding for high model flexibility and expressiveness. Then, a hypercube similarity scoring function is designed to measure the similarity between hypercubes representing users and items. To make a balance between the accuracy and diversity of recommendations and achieve personalized diversity needs, we further develop a user-item pairwise attention mechanism as well as a user uncertainty masking mechanism in LCD-UC. Besides, we present two new metrics for better evaluation on recommendation diversity, which address the issue that existing metrics only consider the coverage of categories while ignore the frequency of categories. The extensive experiments on three real-world datasets show that LCD-UC can improve both recommendation accuracy and diversity over three base models, and is superior to six state-of-the-art recommendation models. An online 10-day AB test also demonstrates that LCD-UC can improve the performance of a real-world advertising system.

The Probability Ranking Principle (PRP) has been considered as the foundational standard in the design of information retrieval (IR) systems. The principle requires an IR module's returned list of results to be ranked with respect to the underlying user interests, so as to maximize the results' utility. Nevertheless, we point out that it is inappropriate to indiscriminately apply PRP through every stage of a contemporary IR system. Such systems contain multiple stages (e.g., retrieval, pre-ranking, ranking, and re-ranking stages, as examined in this paper). The selection bias inherent in the model of each stage significantly influences the results that are ultimately presented to users. To address this issue, we propose an improved ranking principle for multi-stage systems, namely the Generalized Probability Ranking Principle (GPRP), to emphasize both the selection bias in each stage of the system pipeline as well as the underlying interest of users. We realize GPRP via a unified algorithmic framework named Full Stage Learning to Rank. Our core idea is to first estimate the selection bias in the subsequent stages and then learn a ranking model that best complies with the downstream modules' selection bias so as to deliver its top ranked results to the final ranked list in the system's output. We performed extensive experiment evaluations of our developed Full Stage Learning to Rank solution, using both simulations and online A/B tests in one of the leading short-video recommendation platforms. The algorithm is proved to be effective in both retrieval and ranking stages. Since deployed, the algorithm has brought consistent and significant performance gain to the platform.

Contrastive learning has recently emerged as an effective strategy for improving the performance of sequential recommendation. However, traditional models commonly construct the contrastive loss by directly optimizing human-designed positive and negative samples, resulting in a model that is overly sensitive to heuristic rules. To address this limitation, we propose a novel soft contrastive framework for sequential recommendation in this article. Our main idea is to extend the point-wise contrast to a region-level comparison, where we aim to identify instances near the initially selected positive/negative samples that exhibit similar contrastive properties. This extension improves the model’s robustness to human heuristics. To achieve this objective, we introduce an adversarial contrastive loss that allows us to explore the sample regions more effectively. Specifically, we begin by considering the user behavior sequence as a holistic entity. We construct adversarial samples by introducing a continuous perturbation vector to the sequence representation. This perturbation vector adds variability to the sequence, enabling more flexible exploration of the sample regions. Moreover, we extend the aforementioned strategy by applying perturbations directly to the items within the sequence. This accounts for the sequential nature of the items. To capture these sequential relationships, we utilize a recurrent neural network to associate the perturbations, which introduces an inductive bias for more efficient exploration of adversarial samples. To demonstrate the effectiveness of our model, we conduct extensive experiments on five real-world datasets.

Personalized recommender systems fulfill the daily demands of customers and boost online businesses. The goal is to learn a policy that can generate a list of items that matches the user's demand or interest. While most existing methods learn a pointwise scoring model that predicts the ranking score of each individual item, recent research shows that the listwise approach can further improve the recommendation quality by modeling the intra-list correlations of items that are exposed together. This has motivated the recent list reranking and generative recommendation approaches that optimize the overall utility of the entire list. However, it is challenging to explore the combinatorial space of list actions and existing methods that use cross-entropy loss may suffer from low diversity issues. In this work, we aim to learn a policy that can generate sufficiently diverse item lists for users while maintaining high recommendation quality. The proposed solution, GFN4Rec, is a generative method that takes the insight of the flow network to ensure the alignment between list generation probability and its reward. The key advantages of our solution are the log scale reward matching loss that intrinsically improves the generation diversity and the autoregressive item selection model that captures the item mutual influences while capturing future reward of the list. As validation of our method's effectiveness and its superior diversity during active exploration, we conduct experiments on simulated online environments as well as an offline evaluation framework for two real-world datasets.

Life-long user behavior modeling, i.e., extracting a user's hidden interests from rich historical behaviors in months or even years, plays a central role in modern CTR prediction systems. Conventional algorithms mostly follow two cascading stages: a simple General Search Unit (GSU) for fast and coarse search over tens of thousands of long-term behaviors and an Exact Search Unit (ESU) for effective Target Attention (TA) over the small number of finalists from GSU. Although efficient, existing algorithms mostly suffer from a crucial limitation: the inconsistent target-behavior relevance metrics between GSU and ESU. As a result, their GSU usually misses highly relevant behaviors but retrieves ones considered irrelevant by ESU. In such case, the TA in ESU, no matter how attention is allocated, mostly deviates from the real user interests and thus degrades the overall CTR prediction accuracy. To address such inconsistency, we propose TWo-stage Interest Network (TWIN), where our Consistency-Preserved GSU (CP-GSU) adopts the identical target-behavior relevance metric as the TA in ESU, making the two stages twins. Specifically, to break TA's computational bottleneck and extend it from ESU to GSU, or namely from behavior length 102 to length 104 - 105, we build a novel attention mechanism by behavior feature splitting. For the video inherent features of a behavior, we calculate their linear projection by efficient pre-computing & caching strategies. And for the user-item cross features, we compress each into a one-dimentional bias term in the attention score calculation to save the computational cost. The consistency between two stages, together with the effective TA-based relevance metric in CP-GSU, contributes to significant performance gain in CTR prediction. Offline experiments on a 46 billion scale real production dataset from Kuaishou and an Online A/B test show that TWIN outperforms all compared SOTA algorithms. With optimized online infrastructure, we reduce the computational bottleneck by 99.3%, which contributes to the successful deployment of TWIN on Kuaishou, serving the main traffic of hundreds of millions of active users everyday.

With the increase of content pages and interactive buttons in online services such as online-shopping and video-watching websites, industrial-scale recommender systems face challenges in multi-domain and multi-task recommendations. The core of multi-task and multi-domain recommendation is to accurately capture user interests in multiple scenarios given multiple user behaviors. In this paper, we propose a plug-and-play Parameter and Embedding Personalized Network (PEPNet) for multi-domain and multi-task recommendation. PEPNet takes personalized prior information as input and dynamically scales the bottom-level Embedding and top-level DNN hidden units through gate mechanisms. Embedding Personalized Network (EPNet) performs personalized selection on Embedding to fuse features with different importance for different users in multiple domains. Parameter Personalized Network (PPNet) executes personalized modification on DNN parameters to balance targets with different sparsity for different users in multiple tasks. We have made a series of special engineering optimizations combining the Kuaishou training framework and the online deployment environment. By infusing personalized selection of Embedding and personalized modification of DNN parameters, PEPNet tailored to the interests of each individual obtains significant performance gains, with online improvements exceeding 1% in multiple task metrics across multiple domains. We have deployed PEPNet in Kuaishou apps, serving over 300 million users every day.

Graph Neural Networks (GNNs) have demonstrated promising results on exploiting node representations for many downstream tasks through supervised end-to-end training. To deal with the widespread label scarcity issue in real-world applications, Graph Contrastive Learning (GCL) is leveraged to train GNNs with limited or even no labels by maximizing the mutual information between nodes in its augmented views generated from the original graph. However, the distribution of graphs remains unconsidered in view generation, resulting in the ignorance of unseen edges in most existing literature, which is empirically shown to be able to improve GCL's performance in our experiments. To this end, we propose to incorporate graph generative adversarial networks (GANs) to learn the distribution of views for GCL, in order to i) automatically capture the characteristic of graphs for augmentations, and ii) jointly train the graph GAN model and the GCL model. Specifically, we present GACN, a novel Generative Adversarial Contrastive learning Network for graph representation learning. GACN develops a view generator and a view discriminator to generate augmented views automatically in an adversarial style. Then, GACN leverages these views to train a GNN encoder with two carefully designed self-supervised learning losses, including the graph contrastive loss and the Bayesian personalized ranking Loss. Furthermore, we design an optimization framework to train all GACN modules jointly. Extensive experiments on seven real-world datasets show that GACN is able to generate high-quality augmented views for GCL and is superior to twelve state-of-the-art baseline methods. Noticeably, our proposed GACN surprisingly discovers that the generated views in data augmentation finally conform to the well-known preferential attachment rule in online networks.

Current advances in recommender systems have been remarkably successful in optimizing immediate engagement. However, long-term user engagement, a more desirable performance metric, remains difficult to improve. Meanwhile, recent reinforcement learning (RL) algorithms have shown their effectiveness in a variety of long-term goal optimization tasks. For this reason, RL is widely considered as a promising framework for optimizing long-term user engagement in recommendation. Though promising, the application of RL heavily relies on well-designed rewards, but designing rewards related to long-term user engagement is quite difficult. To mitigate the problem, we propose a novel paradigm, recommender systems with human preferences (or Preference-based Recommender systems), which allows RL recommender systems to learn from preferences about users' historical behaviors rather than explicitly defined rewards. Such preferences are easily accessible through techniques such as crowdsourcing, as they do not require any expert knowledge. With PrefRec, we can fully exploit the advantages of RL in optimizing long-term goals, while avoiding complex reward engineering. PrefRec uses the preferences to automatically train a reward function in an end-to-end manner. The reward function is then used to generate learning signals to train the recommendation policy. Furthermore, we design an effective optimization method for PrefRec, which uses an additional value function, expectile regression and reward model pre-training to improve the performance. We conduct experiments on a variety of long-term user engagement optimization tasks. The results show that PrefRec significantly outperforms previous state-of-the-art methods in all the tasks.

Sequential recommendation is one of the most important tasks in recommender systems, which aims to recommend the next interacted item with historical behaviors as input. Traditional sequential recommendation always mainly considers the collected positive feedback such as click, purchase, etc. However, in short-video platforms such as TikTok, video viewing behavior may not always represent positive feedback. Specifically, the videos are played automatically, and users passively receive the recommended videos. In this new scenario, users passively express negative feedback by skipping over videos they do not like, which provides valuable information about their preferences. Different from the negative feedback studied in traditional recommender systems, this passive-negative feedback can reflect users’ interests and serve as an important supervision signal in extracting users’ preferences. Therefore, it is essential to carefully design and utilize it in this novel recommendation scenario. In this work, we first conduct analyses based on a large-scale real-world short-video behavior dataset and illustrate the significance of leveraging passive feedback. We then propose a novel method that deploys the sub-interest encoder, which incorporates positive feedback and passive-negative feedback as supervision signals to learn the user’s current active sub-interest. Moreover, we introduce an adaptive fusion layer to integrate various sub-interests effectively. To enhance the robustness of our model, we then introduce a multi-task learning module to simultaneously optimize two kinds of feedback – passive-negative feedback and traditional randomly-sampled negative feedback. The experiments on two large-scale datasets verify that the proposed method can significantly outperform state-of-the-art approaches. The code is released at https://github.com/tsinghua-fib-lab/RecSys2023-SINE to benefit the community.

Modern online service providers such as online shopping platforms often provide both search and recommendation (S&R) services to meet different user needs. Rarely has there been any effective means of incorporating user behavior data from both S&R services. Most existing approaches either simply treat S&R behaviors separately, or jointly optimize them by aggregating data from both services, ignoring the fact that user intents in S&R can be distinctively different. In our paper, we propose a Search-Enhanced framework for the Sequential Recommendation (SESRec) that leverages users' search interests for recommendation, by disentangling similar and dissimilar representations within S&R behaviors. Specifically, SESRec first aligns query and item embeddings based on users' query-item interactions for the computations of their similarities. Two transformer encoders are used to learn the contextual representations of S&R behaviors independently. Then a contrastive learning task is designed to supervise the disentanglement of similar and dissimilar representations from behavior sequences of S&R. Finally, we extract user interests by the attention mechanism from three perspectives, i.e., the contextual representations, the two separated behaviors containing similar and dissimilar interests. Extensive experiments on both industrial and public datasets demonstrate that SESRec consistently outperforms state-of-the-art models. Empirical studies further validate that SESRec successfully disentangle similar and dissimilar user interests from their S&R behaviors.

Modern recommender systems often deal with a variety of user interactions, e.g., click, forward, purchase, etc., which requires the underlying recommender engines to fully understand and leverage multi-behavior data from users. Despite recent efforts towards making use of heterogeneous data, multi-behavior recommendation still faces great challenges. Firstly, sparse target signals and noisy auxiliary interactions remain an issue. Secondly, existing methods utilizing self-supervised learning (SSL) to tackle the data sparsity neglect the serious optimization imbalance between the SSL task and the target task. Hence, we propose a Multi-Behavior Self-Supervised Learning (MBSSL) framework together with an adaptive optimization method. Specifically, we devise a behavior-aware graph neural network incorporating the self-attention mechanism to capture behavior multiplicity and dependencies. To increase the robustness to data sparsity under the target behavior and noisy interactions from auxiliary behaviors, we propose a novel self-supervised learning paradigm to conduct node self-discrimination at both inter-behavior and intra-behavior levels. In addition, we develop a customized optimization strategy through hybrid manipulation on gradients to adaptively balance the self-supervised learning task and the main supervised recommendation task. Extensive experiments on five real-world datasets demonstrate the consistent improvements obtained by MBSSL over ten state-of-the-art (SOTA) baselines. We release our model implementation at: https://github.com/Scofield666/MBSSL.git.

The wide popularity of short videos on social media poses new opportunities and challenges to optimize recommender systems on the video-sharing platforms. Users sequentially interact with the system and provide complex and multi-faceted responses, including WatchTime and various types of interactions with multiple videos. On the one hand, the platforms aim at optimizing the users’ cumulative WatchTime (main goal) in the long term, which can be effectively optimized by Reinforcement Learning. On the other hand, the platforms also need to satisfy the constraint of accommodating the responses of multiple user interactions (auxiliary goals) such as Like, Follow, Share, etc. In this paper, we formulate the problem of short video recommendation as a Constrained Markov Decision Process (CMDP). We find that traditional constrained reinforcement learning algorithms fail to work well in this setting. We propose a novel two-stage constrained actor-critic method: At stage one, we learn individual policies to optimize each auxiliary signal. In stage two, we learn a policy to (i) optimize the main signal and (ii) stay close to policies learned in the first stage, which effectively guarantees the performance of this main policy on the auxiliaries. Through extensive offline evaluations, we demonstrate the effectiveness of our method over alternatives in both optimizing the main goal as well as balancing the others. We further show the advantage of our method in live experiments of short video recommendations, where it significantly outperforms other baselines in terms of both WatchTime and interactions. Our approach has been fully launched in the production system to optimize user experiences on the platform.

In recommender systems, reinforcement learning solutions have effectively boosted recommendation performance because of their ability to capture long-term user-system interaction. However, the action space of the recommendation policy is a list of items, which could be extremely large with a dynamic candidate item pool. To overcome this challenge, we propose a hyper-actor and critic learning framework where the policy decomposes the item list generation process into a hyper-action inference step and an effect-action selection step. The first step maps the given state space into a vectorized hyper-action space, and the second step selects the item list based on the hyper-action. In order to regulate the discrepancy between the two action spaces, we design an alignment module along with a kernel mapping function for items to ensure inference accuracy and include a supervision module to stabilize the learning process. We build simulated environments on public datasets and empirically show that our framework is superior in recommendation compared to standard RL baselines.

In recent years, Multi-task Learning (MTL) has yielded immense success in Recommender System (RS) applications [40]. However, current MTL-based recommendation models tend to disregard the session-wise patterns of user-item interactions because they are predominantly constructed based on item-wise datasets. Moreover, balancing multiple objectives has always been a challenge in this field, which is typically avoided via linear estimations in existing works. To address these issues, in this paper, we propose a Reinforcement Learning (RL) enhanced MTL framework, namely RMTL, to combine the losses of different recommendation tasks using dynamic weights. To be specific, the RMTL structure can address the two aforementioned issues by (i) constructing an MTL environment from session-wise interactions and (ii) training multi-task actor-critic network structure, which is compatible with most existing MTL-based recommendation models, and (iii) optimizing and fine-tuning the MTL loss function using the weights generated by critic networks. Experiments on two real-world public datasets demonstrate the effectiveness of RMTL with a higher AUC against state-of-the-art MTL-based recommendation models. Additionally, we evaluate and validate RMTL’s compatibility and transferability across various MTL models.

Short video applications have attracted billions of users in recent years, fulfilling their various needs with diverse content. Users usually watch short videos on many topics on mobile devices in a short period of time, and give explicit or implicit feedback very quickly to the short videos they watch. The recommender system needs to perceive users' preferences in real-time in order to satisfy their changing interests. Traditionally, recommender systems deployed at server side return a ranked list of videos for each request from client. Thus it cannot adjust the recommendation results according to the user's real-time feedback before the next request. Due to client-server transmitting latency, it is also unable to make immediate use of users' real-time feedback. However, as users continue to watch videos and feedback, the changing context leads the ranking of the server-side recommendation system inaccurate. In this paper, we propose to deploy a short video recommendation framework on mobile devices to solve these problems. Specifically, we design and deploy a tiny on-device ranking model to enable real-time re-ranking of server-side recommendation results. We improve its prediction accuracy by exploiting users' real-time feedback of watched videos and client-specific real-time features.

Watch-time prediction remains to be a key factor in reinforcing user engagement via video recommendations. It has become increasingly important given the ever-growing popularity of online videos. However, prediction of watch time not only depends on the match between the user and the video but is often mislead by the duration of the video itself. With the goal of improving watch time, recommendation is always biased towards videos with long duration. Models trained on this imbalanced data face the risk of bias amplification, which misguides platforms to over-recommend videos with long duration but overlook the underlying user interests. This paper presents the first work to study duration bias in watch-time prediction for video recommendation. We employ a causal graph illuminating that duration is a confounding factor that concurrently affects video exposure and watch-time prediction---the first effect on video causes the bias issue and should be eliminated, while the second effect on watch time originates from video intrinsic characteristics and should be preserved. To remove the undesired bias but leverage the natural effect, we propose a Duration-Deconfounded Quantile-based (D2Q) watch-time prediction framework, which allows for scalability to perform on industry production systems. Through extensive offline evaluation and live experiments, we showcase the effectiveness of this duration-deconfounding framework by significantly outperforming the state-of-the-art baselines. We have fully launched our approach on Kuaishou App, which has substantially improved real-time video consumption due to more accurate watch-time predictions.

Modern online advertising systems inevitably rely on personalization methods, such as click-through rate (CTR) prediction. Recent progress in CTR prediction enjoys the rich representation capabilities of deep learning and achieves great success in large-scale industrial applications. However, these methods can suffer from lack of exploration. Another line of prior work addresses the exploration-exploitation trade-off problem with contextual bandit methods, which are recently less studied in the industry due to the difficulty in extending their flexibility with deep models. In this paper, we propose a novel Deep Uncertainty-Aware Learning (DUAL) method to learn CTR models based on Gaussian processes, which can provide predictive uncertainty estimations while maintaining the flexibility of deep neural networks. DUAL can be easily implemented on existing models and deployed in real-time systems with minimal extra computational overhead. By linking the predictive uncertainty estimation ability of DUAL to well-known bandit algorithms, we further present DUAL-based Ad-ranking strategies to boost up long-term utilities such as the social welfare in advertising systems. Experimental results on several public datasets demonstrate the effectiveness of our methods. Remarkably, an online A/B test deployed in the Alibaba display advertising platform shows an 8.2% social welfare improvement and an 8.0% revenue lift.

In e-commerce advertising, the ad platform usually relies on auction mechanisms to optimize different performance metrics, such as user experience, advertiser utility, and platform revenue. However, most of the state-of-the-art auction mechanisms only focus on optimizing a single performance metric, e.g., either social welfare or revenue, and are not suitable for e-commerce advertising with various, dynamic, difficult to estimate, and even conflicting performance metrics. In this paper, we propose a new mechanism called Deep GSP auction, which leverages deep learning to design new rank score functions within the celebrated GSP auction framework. These new rank score functions are implemented via deep neural network models under the constraints of monotone allocation and smooth transition. The requirement of monotone allocation ensures Deep GSP auction nice game theoretical properties, while the requirement of smooth transition guarantees the advertiser utilities would not fluctuate too much when the auction mechanism switches among candidate mechanisms to achieve different optimization objectives. We deployed the proposed mechanisms in a leading e-commerce ad platform and conducted comprehensive experimental evaluations with both offline simulations and online A/B tests. The results demonstrated the effectiveness of the Deep GSP auction compared to the state-of-the-art auction mechanisms.

For e-commerce platforms such as Taobao and Amazon, advertisers play an important role in the entire digital ecosystem: their behaviors explicitly influence users' browsing and shopping experience; more importantly, advertiser's expenditure on advertising constitutes a primary source of platform revenue. Therefore, providing better services for advertisers is essential for the long-term prosperity for e-commerce platforms. To achieve this goal, the ad platform needs to have an in-depth understanding of advertisers in terms of both their marketing intents and satisfaction over the advertising performance, based on which further optimization could be carried out to service the advertisers in the correct direction. In this paper, we propose a novel Deep Satisfaction Prediction Network (DSPN), which models advertiser intent and satisfaction simultaneously. It employs a two-stage network structure where advertiser intent vector and satisfaction are jointly learned by considering the features of advertiser's action information and advertising performance indicators. Experiments on an Alibaba advertisement dataset and online evaluations show that our proposed DSPN outperforms state-of-the-art baselines and has stable performance in terms of AUC in the online environment. Further analyses show that DSPN not only predicts advertisers' satisfaction accurately but also learns an explainable advertiser intent, revealing the opportunities to optimize the advertising performance further.

To drive purchase in online advertising, it is of the advertiser's great interest to optimize the sequential advertising strategy whose performance and interpretability are both important. The lack of interpretability in existing deep reinforcement learning methods makes it not easy to understand, diagnose and further optimize the strategy.In this paper, we propose our Deep Intents Sequential Advertising (DISA) method to address these issues. The key part of interpretability is to understand a consumer's purchase intent which is, however, unobservable (called hidden states). In this paper, we model this intention as a latent variable and formulate the problem as a Partially Observable Markov Decision Process (POMDP) where the underlying intents are inferred based on the observable behaviors. Large-scale industrial offline and online experiments demonstrate our method's superior performance over several baselines. The inferred hidden states are analyzed, and the results prove the rationality of our inference.

Rich user behavior data has been proven to be of great value for click-through rate prediction tasks, especially in industrial applications such as recommender systems and online advertising. Both industry and academy have paid much attention to this topic and propose different approaches to modeling with long sequential user behavior data. Among them, memory network based model MIMN proposed by Alibaba, achieves SOTA with the co-design of both learning algorithm and serving system. MIMN is the first industrial solution that can model sequential user behavior data with length scaling up to 1000. However, MIMN fails to precisely capture user interests given a specific candidate item when the length of user behavior sequence increases further, say, by 10 times or more. This challenge exists widely in previously proposed approaches.

Position bias is a critical problem in information retrieval when dealing with implicit yet biased user feedback data. Unbiased ranking methods typically rely on causality models and debias the user feedback through inverse propensity weighting. While practical, these methods still suffer from two major problems. First, when infer a user click, the impact of the contextual information, such as documents that have been examined, is often ignored. Second, only the position bias is considered but other issues resulted from user browsing behaviors are overlooked. In this paper, we propose an end-to-end Deep Recurrent Survival Ranking (DRSR), a unified framework to jointly model user's various behaviors, to (i) consider the rich contextual information in the ranking list; and (ii) address the hidden issues underlying user behaviors, i.e., to mine observe pattern in queries without any click (non-click queries), and to model tracking logs which cannot truly reflect the user browsing intents (untrusted observation). Specifically, we adopt a recurrent neural network to model the contextual information and estimates the conditional likelihood of user feedback at each position. We then incorporate survival analysis techniques with the probability chain rule to mathematically recover the unbiased joint probability of one user's various behaviors. DRSR can be easily incorporated with both point-wise and pair-wise learning objectives. The extensive experiments over two large-scale industrial datasets demonstrate the significant performance gains of our model comparing with the state-of-the-arts.

Most e-commerce product feeds provide blended results of advertised products and recommended products to consumers. The underlying advertising and recommendation platforms share similar if not exactly the same set of candidate products. Consumers' behaviors on the advertised results constitute part of the recommendation model's training data and therefore can influence the recommended results. We refer to this process as Leverage. Considering this mechanism, we propose a novel perspective that advertisers can strategically bid through the advertising platform to optimize their recommended organic traffic. By analyzing the real-world data, we first explain the principles of Leverage mechanism, i.e., the dynamic models of Leverage. Then we introduce a novel Leverage optimization problem and formulate it with a Markov Decision Process. To deal with the sample complexity challenge in model-free reinforcement learning, we propose a novel Hybrid Training Leverage Bidding (HTLB) algorithm which combines the real-world samples and the emulator-generated samples to boost the learning speed and stability. Our offline experiments as well as the results from the online deployment demonstrate the superior performance of our approach.

In E-commerce advertising, where product recommendations and product ads are presented to users simultaneously, the traditional setting is to display ads at fixed positions. However, under such a setting, the advertising system loses the flexibility to control the number and positions of ads, resulting in sub-optimal platform revenue and user experience. Consequently, major e-commerce platforms (e.g., Taobao.com) have begun to consider more flexible ways to display ads. In this paper, we investigate the problem of advertising with adaptive exposure: can we dynamically determine the number and positions of ads for each user visit under certain business constraints so that the platform revenue can be increased? More specifically, we consider two types of constraints: request-level constraint ensures user experience for each user visit, and platform-level constraint controls the overall platform monetization rate. We model this problem as a Constrained Markov Decision Process with per-state constraint (psCMDP) and propose a constrained two-level reinforcement learning approach to decompose the original problem into two relatively independent sub-problems. To accelerate policy learning, we also devise a constrained hindsight experience replay mechanism. Experimental evaluations on industry-scale real-world datasets demonstrate the merits of our approach in both obtaining higher revenue under the constraints and the effectiveness of the constrained hindsight experience replay mechanism.

Click-through rate (CTR) prediction is critical for industrial applications such as recommender system and online advertising. Practically, it plays an important role for CTR modeling in these applications by mining user interest from rich historical behavior data. Driven by the development of deep learning, deep CTR models with ingeniously designed architecture for user interest modeling have been proposed, bringing remarkable improvement of model performance over offline metric. However, great efforts are needed to deploy these complex models to online serving system for realtime inference, facing massive traffic request. Things turn to be more difficult when it comes to long sequential user behavior data, as the system latency and storage cost increase approximately linearly with the length of user behavior sequence.

Real-Time Bidding (RTB) is an important paradigm in display advertising, where advertisers utilize extended information and algorithms served by Demand Side Platforms (DSPs) to improve advertising performance. A common problem for DSPs is to help advertisers gain as much value as possible with budget constraints. However, advertisers would routinely add certain key performance indicator (KPI) constraints that the advertising campaign must meet due to practical reasons. In this paper, we study the common case where advertisers aim to maximize the quantity of conversions, and set cost-per-click (CPC) as a KPI constraint. We convert such a problem into a linear programming problem and leverage the primal-dual method to derive the optimal bidding strategy. To address the applicability issue, we propose a feedback control-based solution and devise the multivariable control system. The empirical study based on real-word data from Taobao.com verifies the effectiveness and superiority of our approach compared with the state of the art in the industry practices.

User response prediction, which models the user preference w.r.t. the presented items, plays a key role in online services. With two-decade rapid development, nowadays the cumulated user behavior sequences on mature Internet service platforms have become extremely long since the user's first registration. Each user not only has intrinsic tastes, but also keeps changing her personal interests during lifetime. Hence, it is challenging to handle such lifelong sequential modeling for each individual user. Existing methodologies for sequential modeling are only capable of dealing with relatively recent user behaviors, which leaves huge space for modeling long-term especially lifelong sequential patterns to facilitate user modeling. Moreover, one user's behavior may be accounted for various previous behaviors within her whole online activity history, i.e., long-term dependency with multi-scale sequential patterns. In order to tackle these challenges, in this paper, we propose a Hierarchical Periodic Memory Network for lifelong sequential modeling with personalized memorization of sequential patterns for each user. The model also adopts a hierarchical and periodical updating mechanism to capture multi-scale sequential patterns of user interests while supporting the evolving user behavior logs. The experimental results over three large-scale real-world datasets have demonstrated the advantages of our proposed model with significant improvement in user response prediction performance against the state-of-the-arts.

In Taobao, the largest e-commerce platform in China, billions of items are provided and typically displayed with their images.For better user experience and business effectiveness, Click Through Rate (CTR) prediction in online advertising system exploits abundant user historical behaviors to identify whether a user is interested in a candidate ad. Enhancing behavior representations with user behavior images will help understand user's visual preference and improve the accuracy of CTR prediction greatly. So we propose to model user preference jointly with user behavior ID features and behavior images. However, training with user behavior images brings tens to hundreds of images in one sample, giving rise to a great challenge in both communication and computation. To handle these challenges, we propose a novel and efficient distributed machine learning paradigm called Advanced Model Server (AMS). With the well-known Parameter Server (PS) framework, each server node handles a separate part of parameters and updates them independently. AMS goes beyond this and is designed to be capable of learning a unified image descriptor model shared by all server nodes which embeds large images into low dimensional high level features before transmitting images to worker nodes. AMS thus dramatically reduces the communication load and enables the arduous joint training process. Based on AMS, the methods of effectively combining the images and ID features are carefully studied, and then we propose a Deep Image CTR Model. Our approach is shown to achieve significant improvements in both online and offline evaluations, and has been deployed in Taobao display advertising system serving the main traffic.

Real-time advertising allows advertisers to bid for each impression for a visiting user. To optimize specific goals such as maximizing revenue and return on investment (ROI) led by ad placements, advertisers not only need to estimate the relevance between the ads and user's interests, but most importantly require a strategic response with respect to other advertisers bidding in the market. In this paper, we formulate bidding optimization with multi-agent reinforcement learning. To deal with a large number of advertisers, we propose a clustering method and assign each cluster with a strategic bidding agent. A practical Distributed Coordinated Multi-Agent Bidding (DCMAB) has been proposed and implemented to balance the tradeoff between the competition and cooperation among advertisers. The empirical study on our industry-scaled real-world data has demonstrated the effectiveness of our methods. Our results show cluster-based bidding would largely outperform single-agent and bandit approaches, and the coordinated bidding achieves better overall objectives than purely self-interested bidding agents.

Real-time bidding (RTB) is an important mechanism in online display advertising, where a proper bid for each page view plays an essential role for good marketing results. Budget constrained bidding is a typical scenario in RTB where the advertisers hope to maximize the total value of the winning impressions under a pre-set budget constraint. However, the optimal bidding strategy is hard to be derived due to the complexity and volatility of the auction environment. To address these challenges, in this paper, we formulate budget constrained bidding as a Markov Decision Process and propose a model-free reinforcement learning framework to resolve the optimization problem. Our analysis shows that the immediate reward from environment is misleading under a critical resource constraint. Therefore, we innovate a reward function design methodology for the reinforcement learning problems with constraints. Based on the new reward design, we employ a deep neural network to learn the appropriate reward so that the optimal policy can be learned effectively. Different from the prior model-based work, which suffers from the scalability problem, our framework is easy to be deployed in large-scale industrial applications. The experimental evaluations demonstrate the effectiveness of our framework on large-scale real datasets.

Click-through rate prediction is an essential task in industrial applications, such as online advertising. Recently deep learning based models have been proposed, which follow a similar Embedding&MLP paradigm. In these methods large scale sparse input features are first mapped into low dimensional embedding vectors, and then transformed into fixed-length vectors in a group-wise manner, finally concatenated together to fed into a multilayer perceptron (MLP) to learn the nonlinear relations among features. In this way, user features are compressed into a fixed-length representation vector, in regardless of what candidate ads are. The use of fixed-length vector will be a bottleneck, which brings difficulty for Embedding&MLP methods to capture user's diverse interests effectively from rich historical behaviors. In this paper, we propose a novel model: Deep Interest Network (DIN) which tackles this challenge by designing a local activation unit to adaptively learn the representation of user interests from historical behaviors with respect to a certain ad. This representation vector varies over different ads, improving the expressive ability of model greatly. Besides, we develop two techniques: mini-batch aware regularization and data adaptive activation function which can help training industrial deep networks with hundreds of millions of parameters. Experiments on two public datasets as well as an Alibaba real production dataset with over 2 billion samples demonstrate the effectiveness of proposed approaches, which achieve superior performance compared with state-of-the-art methods. DIN now has been successfully deployed in the online display advertising system in Alibaba, serving the main traffic.

Model-based methods for recommender systems have been studied extensively in recent years. In systems with large corpus, however, the calculation cost for the learnt model to predict all user-item preferences is tremendous, which makes full corpus retrieval extremely difficult. To overcome the calculation barriers, models such as matrix factorization resort to inner product form (i.e., model user-item preference as the inner product of user, item latent factors) and indexes to facilitate efficient approximate k-nearest neighbor searches. However, it still remains challenging to incorporate more expressive interaction forms between user and item features, e.g., interactions through deep neural networks, because of the calculation cost. In this paper, we focus on the problem of introducing arbitrary advanced models to recommender systems with large corpus. We propose a novel tree-based method which can provide logarithmic complexity w.r.t. corpus size even with more expressive models such as deep neural networks. Our main idea is to predict user interests from coarse to fine by traversing tree nodes in a top-down fashion and making decisions for each user-node pair. We also show that the tree structure can be jointly learnt towards better compatibility with users' interest distribution and hence facilitate both training and prediction. Experimental evaluations with two large-scale real-world datasets show that the proposed method significantly outperforms traditional methods. Online A/B test results in Taobao display advertising platform also demonstrate the effectiveness of the proposed method in production environments.

Estimating post-click conversion rate (CVR) accurately is crucial for ranking systems in industrial applications such as recommendation and advertising. Conventional CVR modeling applies popular deep learning methods and achieves state-of-the-art performance. However it encounters several task-specific problems in practice, making CVR modeling challenging. For example, conventional CVR models are trained with samples of clicked impressions while utilized to make inference on the entire space with samples of all impressions. This causes a sample selection bias problem. Besides, there exists an extreme data sparsity problem, making the model fitting rather difficult. In this paper, we model CVR in a brand-new perspective by making good use of sequential pattern of user actions, i.e., impression -> click -> conversion. The proposed Entire Space Multi-task Model (ESMM) can eliminate the two problems simultaneously by i) modeling CVR directly over the entire space, ii) employing a feature representation transfer learning strategy. Experiments on dataset gathered from Taobao's recommender system demonstrate that ESMM significantly outperforms competitive methods. We also release a sampling version of this dataset to enable future research. To the best of our knowledge, this is the first public dataset which contains samples with sequential dependence of click and conversion labels for CVR modeling.

Taobao, as the largest online retail platform in the world, provides billions of online display advertising impressions for millions of advertisers every day. For commercial purposes, the advertisers bid for specific spots and target crowds to compete for business traffic. The platform chooses the most suitable ads to display in tens of milliseconds. Common pricing methods include cost per mille (CPM) and cost per click (CPC). Traditional advertising systems target certain traits of users and ad placements with fixed bids, essentially regarded as coarse-grained matching of bid and traffic quality. However, the fixed bids set by the advertisers competing for different quality requests cannot fully optimize the advertisers' key requirements. Moreover, the platform has to be responsible for the business revenue and user experience. Thus, we proposed a bid optimizing strategy called optimized cost per click (OCPC) which automatically adjusts the bid to achieve finer matching of bid and traffic quality of page view (PV) request granularity. Our approach optimizes advertisers' demands, platform business revenue and user experience and as a whole improves traffic allocation efficiency. We have validated our approach in Taobao display advertising system in production. The online A/B test shows our algorithm yields substantially better results than previous fixed bid manner.