Dwell time (DT) is a critical post-click metric for evaluating user preference in recommender systems, complementing the traditional click-through rate (CTR). Although multi-task learning is widely adopted to jointly optimize DT and CTR, we observe that multi-task models systematically collapse their DT predictions to the shortest and longest bins, under-predicting the moderate durations. We attribute this moderate-duration bin under-representation to over-reliance on the CTR-DT spurious correlation, and propose ORCA to address it with causal-decoupling. Specifically, ORCA explicitly models and subtracts CTR's negative transfer while preserving its positive transfer. We further introduce (i) feature-level counterfactual intervention, and (ii) a task-interaction module with instance inverse-weighting, weakening CTR-mediated effect and restoring direct DT semantics. ORCA is model-agnostic and easy to deploy. Experiments show an average 10.6% lift in DT metrics without harming CTR. Code is available at https://github.com/Chrissie-Law/ORCA-Mitigating-Over-Reliance-for-Multi-Task-Dwell-Time-Prediction-with-Causal-Decoupling.

Multi-modal recommender systems (MRSs) are pivotal in diverse online web platforms and have garnered considerable attention in recent years. However, previous studies overlook the challenges of (1)noisy multi-modal content, (2) noisy user feedback, and (3) aligning multi-modal content and user feedback. To tackle these challenges, we propose Denoising and Aligning Multi-modal Recommender System (DA-MRS). To mitigate noise in multi-modal content, DA-MRS first constructs item-item graphs determined by consistent content similarity across modalities. To denoise user feedback, DA-MRS associates the probability of observed feedback with multi-modal content and devises a denoised BPR loss. Furthermore, DA-MRS implements Alignment guided by User preference to enhance task-specific item representation and Alignment guided by graded Item relations to provide finer-grained alignment. Extensive experiments verify that DA-MRS is a plug-and-play framework and achieves significant and consistent improvements across various datasets, backbone models, and noisy scenarios.

Recommendation systems, widely adopted in social networks, personalize user experiences through advanced technologies such as Reinforcement Learning (RL), known for producing high-performance, list- wise recommendations. However, RL-based recommendation methods exhibit biases, specifically: 1) Online bias, which stems from a complex real-worldonline policycomposed of various rules and models rather than a single policy; 2) Training bias, a distributional shift resulting from differences between thetarget policyand thebehavior policy. To address these issues, we introduce a novel framework named Contrastive Imitation Proximal Policy Optimization (CIPPO) for recommendation based on RL. This approach leverages extensively labeled feedback data and incorporates a Masked Imitation Network (MIN) that closely emulates the online policy, thus reducing discrepancies between online and offline environments. Additionally, the clipping function in Proximal Policy Optimization, combined with a specially designed contrastive module, effectively reduces the distributional shift between the behavior and target policies. We conduct offline and online experiments to show the improvements of CIPPO, providing details including ablation tests and parameter analysis to validate the effectiveness and robustness. CIPPO gains 12.79% on ACN and in WeChat Top Stories, a large media platform with over 50 million users.

The emergence of Graph Neural Networks (GNNs) has greatly advanced the development of recommendation systems. Recently, many researchers have leveraged GNN-based models to learn fair representations for users and items. However, current GNN-based models suffer from biased user–item interaction data, which negatively impacts recommendation fairness. Although there have been several studies employing adversarial learning to mitigate this issue in recommendation systems, they mostly focus on modifying the model training approach with fairness regularization and neglect direct intervention of biased interaction. In contrast to these models, this article introduces a novel perspective by directly intervening in observed interactions to generate a counterfactual graph (called FairGap) that is not influenced by sensitive node attributes, enabling us to learn fair representations for users and items easily. We design FairGap to answer the key counterfactual question: “Would interactions with an item remain unchanged if a user’s sensitive attributes were concealed?”. We also provide theoretical proofs to show that our learning strategy via the counterfactual graph is unbiased in expectation. Moreover, we propose a fairness-enhancing mechanism to continuously improve user fairness in the graph-based recommendation. Extensive experimental results against state-of-the-art competitors and base models on three real-world datasets validate the effectiveness of our proposed model.

There are roughly three stages in real industrial recommendation systems, candidates generation (retrieval), ranking and reranking. Both individual-level diversity and system-level diversity are important in this framework. The former focus on each single user's experience, while the latter focus on the difference among users. Graph-based retrieval strategies are inevitably hijacked by heavy users and popular items, leading to similar results for each user and the convergence among users. In the reranking phase, Determinantal Point Process (DPP) is widely deployed to increase individual-level diversity. Heavily relying on the semantic information of items, DPP suffers from clickbait and inaccurate categories. Besides, most studies only focus on one of the two levels of diversity, and ignore the mutual influence among different stages in real recommender systems. We argue that individual-level diversity and system-level diversity should be viewed as an integrated problem, and we provide an efficient and deployable solution for web-scale recommenders. Generally, we propose to employ the retrieval graph information in diversity-based reranking, by which to weaken the hidden similarity of items exposed to users, and consequently gain more graph explorations to improve the system-level diversity. Besides, we argue that users' propensity for diversity changes over time in content feed recommendation. Therefore, with the explored graph, we also propose to capture the user's real-time personalized propensity to the diversity. We implement and deploy the combined system in WeChat App's Top Stories used by hundreds of millions of users. Comparative and long-term online A/B tests show our solution can effectively improve both user engagement and system revenue.

Personalized recommendation aims to provide appropriate items according to user preferences mainly from their behaviors. Excessive homogeneous user behaviors on similar items will lead to fatigue, which may decrease user activeness and degrade user experience. However, existing models seldom consider user fatigue in recommender systems. In this work, we propose a novel multi-granularity fatigue, modeling user fatigue from coarse to fine. Specifically, we focus on the recommendation feed scenario, where the underexplored global session fatigue and coarse-grained taxonomy fatigue have large impacts. We conduct extensive analyses to demonstrate the characteristics and influence of different types of fatigues in real-world recommender systems. In experiments, we verify the effectiveness of multi-granularity fatigue in both offline and online evaluations. Currently, the fatigue-enhanced model has also been deployed on a widely-used recommendation system of WeChat.

An effective online recommendation system should jointly capture users' long-term and short-term preferences in both users' internal behaviors (from the target recommendation task) and external behaviors (from other tasks). However, it is extremely challenging to conduct fast adaptations to real-time new trends while making full use of all historical behaviors in large-scale systems, due to the real-world limitations in real-time training efficiency and external behavior acquisition. To address these practical challenges, we propose a novel Long Short-Term Temporal Meta-learning framework (LSTTM) for online recommendation. It arranges user multi-source behaviors in a global long-term graph and an internal short-term graph, and conducts different GAT-based aggregators and training strategies to learn user short-term and long-term preferences separately. To timely capture users' real-time interests, we propose a temporal meta-learning method based on MAML under an asynchronous optimization strategy for fast adaptation, which regards recommendations at different time periods as different tasks. In experiments, LSTTM achieves significant improvements on both offline and online evaluations. It has been deployed on a widely-used online recommendation system named WeChat Top Stories, affecting millions of users.

Personalized recommendation often relies on user historical behaviors to provide items for users. It is intuitive that future information also contains essential messages as supplements to user historical behaviors. However, we cannot directly encode future information into models, since we are unable to get future information in online serving. In this work, we propose a novel adversarial future encoding (AFE) framework to make full use of informative future features in different types of recommendation models. Specifically, AFE contains a future-aware discriminator and a generator. The future-aware discriminator takes both common features and future features as inputs, working as a recommendation prophet to judge user-item pairs. In contrast, the generator is considered as a challenger, which generates items with only common features, aiming to confuse the future-aware prophet. The future-aware discriminator can inspire the generator (to be deployed online) to produce better results. We further conduct a multi-factor optimization to enable a fast and stable model convergence via the direct learning and knowledge distillation losses. Moreover, we have adopted AFE on both a list-wise RL-based ranking model and a point-wise ranking model to verify its universality. In experiments, we conduct sufficient evaluations on two large-scale datasets, achieving significant improvements on both offline and online evaluations. Currently, we have deployed AFE on a real-world system, affecting millions of users. The source code is in https://github.com/modriczhang/AFE.

Reinforcement learning (RL) has been verified in real-world list-wise recommendation. However, RL-based recommendation suffers from huge memory and computation costs due to its large-scale models. Knowledge distillation (KD) is an effective approach for model compression widely used in practice. However, RL-based models strongly rely on sufficient explorations on the enormous user-item space due to the data sparsity issue, which multiplies the challenges of KD with RL models. What the teacher should teach and how much the student should learn from each lesson need to be carefully designed. In this work, we propose a novel Distilled reinforcement learning framework for recommendation (DRL-Rec), which aims to improve both effectiveness and efficiency in list-wise recommendation. Specifically, we propose an Exploring and filtering module before the distillation, which decides what lessons the teacher should teach from both teachers' and students' aspects. We also conduct a Confidence-guided distillation at both output and intermediate levels with a list-wise KL divergence loss and a Hint loss, which aims to understand how much the student should learn for each lesson. We achieve significant improvements on both offline and online evaluations in a well-known recommendation system. DRL-Rec has been deployed on WeChat Top Stories for more than six months, affecting millions of users. The source codes are released in https://github.com/modriczhang/DRL-Rec.

Users of recommendation systems usually focus on one topic at a time. When finishing reading an item, users may want to access more relevant items related to the last read one as extended reading. However, conventional recommendation systems are hard to provide the continuous extended reading function of these relevant items, since the main recommendation results should be diversified. In this paper, we propose a new task named recommendation suggestion, which aims to (1) predict whether users want extended reading, and (2) provide appropriate relevant items as suggestions. These recommended relevant items are arranged in a relevant box and instantly inserted below the clicked item in the main feed. The challenge of recommendation suggestion on relevant items is that it should further consider semantic relevance and information gain besides CTR-related factors. Moreover, the real-time relevant box insertion may also harm the overall performance when users do not want extended reading. To address these issues, we propose a novel Real-time relevant recommendation suggestion (R3S) framework, which consists of an Item recommender and a Box trigger. We extract features from multiple aspects including feature interaction, semantic similarity and information gain as different experts, and propose a new Multi-critic multi-gate mixture-of-experts (M3oE) strategy to jointly consider different experts with multi-head critics. In experiments, we conduct both offline and online evaluations on a real-world recommendation system with detailed ablation tests. The significant improvements in item/box related metrics verify the effectiveness of R3S. Moreover, we have deployed R3S on WeChat Top Stories, which affects millions of users. The source codes are in https://github.com/modriczhang/R3S.

Real-world recommendation systems usually have different learning objectives and evaluation criteria on accuracy, diversity or novelty. Therefore, multi-objective recommendation (MOR) has been widely explored to jointly model different objectives. Pareto efficiency, where no objective can be further improved without hurting others, is viewed as an optimal situation in multi-objective optimization. Recently, Pareto efficiency model has been introduced to MOR, while all existing scalarization methods only have shared objective weights for all instances. To capture users’ objective-level preferences and enhance personalization in Pareto-efficient recommendation, we propose a novel Personalized Approximate Pareto-Efficient Recommendation (PAPERec) framework for multi-objective recommendation. Specifically, we design an approximate Pareto-efficient learning based on scalarization with KKT conditions that closely mimics Pareto efficiency, where users have personalized weights on different objectives. We propose a Pareto-oriented reinforcement learning module to find appropriate personalized objective weights for each user, with the weighted sum of multiple objectives’ gradients considered in reward. In experiments, we conduct extensive offline and online evaluations on a real-world recommendation system. The significant improvements verify the effectiveness of PAPERec in practice. We have deployed PAPERec on WeChat Top Stories, affecting millions of users. The source codes are released in https://github.com/onepunch-cyber/PAPERec.

In article recommendation, the amount of time user spends on viewing articles, dwell time, is an important metric to measure the post-click engagement of user on content and has been widely used as a proxy to user satisfaction, complementing the click feedback. Recently, the sequential pattern of impression-click-read has become one of the most popular type of article recommendation service in real world, where users are presented with a list of titles at first, then get interested in one and click in for reading. Predicting dwell time in such service is conditioned on the click, since the user reads the article only after he clicks the corresponding title. We argue that conventional models for dwell time prediction, which mainly focus on the relevance between the content and the general preference of user, are not well-designed for such service. There is a natural assumption in recommendation system that the click indicates user's getting attracted by the item. Therefore, in the pattern of impression-click-read, the user might get interested and curious on some other concepts different from his general preference while reading, due to the attraction of the title. Conventional models tend to ignore the gap between such temporary interest and the general preference of user in the reading behavior, which fails to use the pattern of impression-click-read and the assumption of the click very well. In this work, we propose a framework, Click-guide Network (CGN) for dwell time prediction, which makes good use of the sequential pattern and the assumption to model the ”guidance” of the click on user preference. CGN is a joint learner for dwell time and click through rate (CTR). We introduce the CTR task as an auxiliary task to help us better learn the preference of user and the representation of title. Besides, we propose the Guider to capture the user's temporary interest raised by the title. We collect the data from WeChat, a widely-used mobile app in China, for experiments. The results demonstrate the advantages of CGN over several competitive baselines on dwell time prediction, while our case studies show how the Guider effectively capture the temporary interest of user.

Cluster-of-bandit policy leverages contextual bandits in a collaborative filtering manner and aids personalized services in the online recommendation system (RecSys). When facing insufficient observations, the cluster-of-bandit policy could achieve more outstanding performance because of knowledge sharing. Cluster-of-bandit policy aims to maximize the cumulative feedback, e.g., clicks, from users. Nevertheless, in the way of their goal exist two kinds of uncertainties. First, cluster-of-bandit algorithms make recommendations according to their uncertain estimation of user interests. Second, cluster-of-bandit algorithms transfer relevant knowledge upon uncertain and noisy user clusters. Existing algorithms only consider the first one, while leaving the latter one untouched. To address the two challenges together, in this paper, we propose the ClexB policy for online RecSys. On the one hand, ClexB estimates user clustering more accurately and with less uncertainty via explorable-clustering. On the other hand, ClexB also exploits and explores user interests by sharing information within and among user clusters. In summary, ClexB explores knowledge transfer and further aids the inferences about user interests. Besides, we provide extensive empirical experiments on both the synthetic and real-world datasets and regret analysis, further consolidating the superiority of ClexB.

Online news streaming services have been one of the major information acquisition resources for mobile users. In many cases, users click an article but find it cannot satisfy or even annoy them. Intuitively, these negative experiences will affect users' behaviors and satisfaction, but such effects have not been well understood. In this work, a retrospective analysis is conducted using real users' log data, containing user's explicit feedback of negative experiences, from a commercial news streaming application. Through multiple intra-session comparison experiments, we find that in current session, users will spend less time reading the content, lose activeness and leave sooner after having negative experiences. Later return and significant changes of user behaviors in the next session are also observed, which demonstrates the existence of inter-session effects of negative experiences.