Online platforms such as Amazon and Netflix serve users across multiple countries and regions, underscoring the importance of multi-market recommendation (MMR). Most MMR methods adopt a pre-training and fine-tuning paradigm, in which a unified model is first trained on centralized, global data and subsequently adapted to specific markets. However, this approach ignores the privacy of market data. While traditional federated learning preserves privacy, it typically aims to obtain a global model by aggregating model parameters and does not account for significant market heterogeneity. Additionally, because ID spaces are disjoint across markets, embedding-based aggregation strategies become ineffective. To overcome these challenges, we propose a federated collaborative signal quantization (FedMM) method for multi-market click-through rate (CTR) prediction. Our core idea leverages a discrete codebook mechanism to achieve privacy-preserving transmission and align disjoint ID spaces. We further employ a hierarchical codebook structure to capture cross-market shared patterns and market-specific characteristics. Specifically, we deploy a residual quantized variational autoencoder (RQ-VAE) with a dual-layer codebook mechanism for each market to quantize collaborative embeddings. The first layer utilizes a global federated codebook, updated via aggregation to capture universally shared collaborative patterns, while the second layer maintains a local codebook to learn market-specific semantics. Finally, the learned discrete codes, which integrate both general and specific collaborative signals, are incorporated into downstream CTR models to enhance prediction accuracy across all markets. Extensive experiments on benchmark datasets demonstrate that FedMM significantly improves recommendation performance with privacy guarantees.

Multi-scenario multi-task recommendation (MSMTR) systems must address recommendation demands across diverse scenarios while simultaneously optimizing multiple objectives, such as click-through rate and conversion rate. Existing MSMTR models typically consist of four information units: scenario-shared, scenario-specific, task-shared, and task-specific networks. These units interact to generate four types of relationship information flows, directed from scenario-shared or scenario-specific networks to task-shared or task-specific networks. However, these models face two main limitations: 1) They often rely on complex architectures, such as mixture-of-experts (MoE) networks, which increase the complexity of information fusion, model size, and training cost. 2) They extract all available information flows without filtering out irrelevant or even harmful content, introducing potential noise. Regarding these challenges, we propose a lightweight Automated Information Flow Selection (AutoIFS) framework for MSMTR. To tackle the first issue, AutoIFS incorporates low-rank adaptation (LoRA) to decouple the four information units, enabling more flexible and efficient information fusion with minimal parameter overhead. To address the second issue, AutoIFS introduces an information flow selection network that automatically filters out invalid scenario-task information flows based on model performance feedback. It employs a simple yet effective pruning function to eliminate useless information flows, thereby enhancing the impact of key relationships and improving model performance. Finally, we evaluate AutoIFS and confirm its effectiveness through extensive experiments on two public benchmark datasets and an online A/B test.

In recent years, more and more researchers and practitioners have focused on multi-domain CTR prediction and achieved great success. Though users’ behaviors often exhibit sequentiality, little effort has been made on multi-domain sequential recommendation (MDSR). Most existing works on MDSR sort the interactions from all domains in chronological order and then predict the next interacted items in each domain. However, they neglect separate interaction sequences in each domain. Therefore, they cannot exploit the commonalities and differences among different domains well. Cross-domain sequential recommendation (CDSR) models are usually designed for performance improvement in one target domain rather than in each domain. Although extending a CDSR model to an MDSR one directly or indirectly is feasible, it will result in high time complexity. Meanwhile, they often ignore data imbalance across different domains, which might cause negative transfer. As a response, we propose a novel MDSR solution called multi-sequence multi-task learning (MML). Our MML consists of three modules, including hybrid-domain sequential preference learning (HSPL), intra-domain sequential preference learning (ISPL) and multi-task learning & prediction (MLP). Specifically, HSPL aims to learn hybrid-domain sequential preferences. Meanwhile, we construct augmented sequences and leverage contrastive learning to learn more unbiased hybrid-domain sequential preferences for alleviating negative transfer. ISPL is designed to capture intra-domain sequential preferences. In the MLP module, three specific tasks and a behavior regularizer are leveraged to ensure that each module can learn the corresponding preferences sufficiently and enhance knowledge transfer among different domains. We conduct extensive experiments on some public datasets using different backbone models and show that our MML is able to achieve significantly better performance than the state-of-the-art methods in two or more domains in most cases. Meanwhile, our MML can achieve the same time complexity as the MDSR models only using hybrid interaction sequences. The source code can be found at https://github.com/plw2019/MML .

While previous works on Multi-Behavior Sequential Recommendation (MBSR) have made great efforts to exploit the behavior types, they often fail to fully leverage the behavior and attribute information to learn different aspects of user preferences. Firstly, for predicting user preferences under target behaviors, there is considerable room for improvement by capturing both intra- and inter-behavior item transition relationships at the sequence level, as well as extracting the inter-behavior common and different aspects of a user’s preferences at each timestep based on the contextual sequential information. Secondly, in MBSR, the effect of behavior types on item attributes has not been revealed, which is very important in real-world applications since a behavior such as view representing a user’s weak preferences could not explicitly indicate the user’s interests on an attribute. To address the above two issues, we propose a novel solution called Behavior-aware Attribute-infused Sequential Encoders (BASE) . Specifically, a module named tri-path preference learning is designed to capture user multi-aspect interests via a sequence encoder, a behavior encoder and an attribute encoder. Besides, a behavior-aware dual-granularity contrastive learning module constructs sample pairs in item and attribute granularities to enhance the model’s ability to utilize potential item and attribute information. Extensive empirical studies on three public datasets show that our BASE significantly outperforms various state-of-the-art methods from five different categories. The datasets and our implementation codes are released at https://github.com/Erin-Gr/BASE .

Multi-Behavior Recommendation (MBR) aims to improve the performance w.r.t. the target behavior (i.e., purchase) by leveraging auxiliary behaviors (e.g., click, favorite). However, in real-world scenarios, a recommendation method often needs to process different types of behaviors and generate personalized lists for each task (i.e., each behavior type). Such a new recommendation problem is referred to as Multi-Behavior Multi-Task Recommendation (MMR). So far, the most powerful MBR methods usually model multi-behavior interactions using a cascading graph paradigm. Although significant progress has been made in optimizing the performance of the target behavior, it often neglects the performance of auxiliary behaviors. To compensate for the deficiencies of the cascading paradigm, we propose a novel solution for MMR, i.e., Behavior-Informed Graph Embedding Learning (BiGEL). Specifically, we first obtain a set of behavior-aware embeddings by using a cascading graph paradigm. Subsequently, we introduce three key modules to improve the performance of the model. The Cascading Gated Feedback (CGF) module enables a feedback-driven optimization process by integrating feedback from the target behavior to refine the auxiliary behaviors preferences. The Global Context Enhancement (GCE) module integrates the global context to maintain the user’s overall preferences, preventing the loss of key preferences due to individual behavior graph modeling. Finally, the Contrastive Preference Alignment (CPA) module addresses the potential changes in user preferences during the cascading process by aligning the preferences of the target behaviors with the global preferences through contrastive learning. Extensive experiments on two real-world datasets demonstrate the effectiveness of our BiGEL compared with 12 very competitive methods. The datasets, scripts, and source code are available at https://github.com/Wenhaocarry/BiGEL .

Most existing GNN-based recommendation methods focus on exploiting a user–item heterogeneous graph, which, however, will cause efficiency and effectiveness challenges, in a federated learning setting considering user privacy. We find that a user–user or item–item homogeneous graph is often privacy-insensitive and can significantly enhance the efficiency and effectiveness of federated graph embedding learning. Hence, we propose a novel framework called Federated Homogeneous Graph Neural Network (FedHoG) , which can provide privacy-preserving recommendations with high-quality and communication-efficient graph learning. We first design a privacy-preserving homogeneous graph construction method, which enables the server to construct an item–item graph and a user–user graph without leaking user privacy. Then, we develop a federated homogeneous graph learning method that enables balanced GNN model training among the server and clients. We also propose a lightweight homogeneous graph convolution method to achieve better graph embedding learning. Finally, extensive experiments on three public datasets show the advantages of our FedHoG in performance and efficiency. The datasets, source codes, and scripts are available at https://github.com/XZHhong/FedHoG .

Multi-behavior sequential recommendation (MBSR), which captures sequential patterns and behavioral heterogeneity to model users’ multifaceted preferences, has shown promising results. Despite their effectiveness, existing methods often suffer from performance degradation due to inherent data sparsity in real-world scenarios. Current data augmentation methods in recommendation systems predominantly focus on single-behavior modeling, failing to account for the diversity of user preference expressions across different types of behaviors. Moreover, conventional augmentation strategies risk introducing noise or irrelevant patterns during sample generation, potentially distorting the next-item prediction task. To address these challenges, we propose a novel and generic framework called multi-behavior data augmentation for sequential recommendation (MBASR). Specifically, we propose five distinct behavior-aware data augmentation operations, which are designed based on interactions both within and across subsequences, to generate diverse and enriched training samples. Each augmentation operation leverages correlations between behaviors or similarities among users, ensuring that the enhanced data remains aligned with users’ natural behavior patterns. Furthermore, we introduce a combined augmentation method, merging two data augmentation operations to achieve better results. In addition, we introduce two position-based sampling strategy that can effectively reduce the perturbation brought by the augmentation operations to the original data. Notably, as a data-centric solution, our MBASR can be seamlessly integrated into various MBSR models without modifying their underlying structures. Comprehensive evaluations on four real-world datasets validate the efficacy of our MBASR, demonstrating significant performance improvement across mainstream MBSR models. The source code, scripts and datasets are available at https://github.com/XiaoQi-C/MBASR

Sequential recommendation (SR) aims to capture users' dynamic preferences based on their historical interactions and provide personalized next-item prediction. Multi-behavior SR (MBSR) further considers behavior types of user-item interactions, which can reveal diverse user interests and alleviate the data sparsity issue w.r.t. the target purchase behaviors. Most existing MBSR approaches ignore the importance of target items closely related to user interests. Moreover, they often suffer from the problem of limited vector representation capability. To tackle the above two challenges, we propose a novel solution called target item-oriented conditional diffusion differential Transformer (ICDDT). Specifically, our ICDDT introduces distribution representations via the diffusion model, allowing effective utilization of target item information during training to better capture user preferences. Firstly, our ICDDT achieves a more appropriate behavior-aware step selection in the diffusion phase by distinguishing the sampling distributions of diffusion steps w.r.t. behavior types. Secondly, our ICDDT introduces three conditions of interaction sequences, target behaviors and diffusion steps into the reverse phase to guide the training of the differential Transformer-based approximator, generating denoised target item representations as user personalized interests. Finally, our ICDDT sets an inference step truncation factor to fit the diffusion step sampling distributions and accelerate the inference process. We conduct extensive experiments on two real-world datasets, where the results show that our ICDDT significantly outperforms all baselines on all metrics. The datasets, source codes and scripts are available at https://github.com/Erin-Gr/ICDDT.

Multi-scenario recommendation (MSR) is a popular training paradigm in industrial platforms for uniformly integrating information from multiple scenarios and serving them simultaneously. A key challenge in MSR research is accurately identifying the commonalities and distinctive information between scenarios. Currently, most existing MSR methods focus on implicitly extracting this information from the architectural level. However, this continues to increase the complexity and training overhead of MSR. Furthermore, the custom components responsible for extracting implicit information in each MSR method are too dependent on the specific MSR architecture and are not easily reused in other methods. Given these challenges, we first show in a motivating experiment that it may be beneficial to explicitly select a reasonable set of shared instances that can affect parameter optimization in all scenarios during the training of MSR, i.e., to explicitly obtain the critical information required for MSR from the data level. Then, this paper proposes SSIM with an adaptive selection network. Specifically, SSIM can be integrated with existing MSR methods in a lightweight way to adaptively select an informative and shareable subset of instances from each scenario to improve recommendations. In particular, the selected multi-scenario shared subset has extraordinary reusability and can be easily saved to benefit model training of various future MSR models. Finally, we evaluate SSIM and demonstrate its effectiveness through experiments on two public multi-scenario benchmarks and an online A/B test.

Sequential recommendation (SR) aims to infer users' future interests and suggest the next items for them. Most SR methods learn one single vector to represent a user's recent interests, i.e., a user representation. Despite their great success, most of them do not explicitly consider the phenomenon of users' interest drift when learning user representations. Moreover, interest drift presents two critical challenges for these SR methods: (1) how to explore the potential distributions of the users' varying interest drift levels; and (2) how to capture the interest drift-aware collaborative knowledge among the users. In this paper, we delve into the issue of interest drift in SR and propose a novel and generic framework, i.e., Interest Drift-driven User Representation Learning (IDURL), to enhance SR methods to tackle the above two challenges. Specifically, our IDURL contains an interest drift quantization (IDQ) module to enable a quantitative measurement of the interest drift. Moreover, a drift representation generation module models the users' latent varying levels of interest drift, and an interest drift-guided representation disentanglement module optimizes the distributions of the interest drift levels under the guidance of IDQ. Furthermore, an interest drift-aware representation alignment module helps to capture the interest drift-aware collaborative knowledge among users. Finally, the users' overall interest representations are obtained to calculate the preference scores on the candidate items. Extensive experiments on four public datasets show the effectiveness of our IDURL. The source code of our IDURL is available at: https://github.com/xiaolLIN/IDURL.

Multi-scenario recommendation (MSR) has become a core component of various online platforms, but its increasing model size has also brought attention to its efficiency optimization. An important effort is to find effective and efficient feature embedding layers for MSR, and existing work focuses on scenario-level feature selection, i.e., all instance embeddings in the same scenario get the same filtering results on the feature set, and the filtering results are different for different scenarios. However, this ignores the information redundancy of the dimension set and the individuality of different instances in the same scenario. To address these limitations, we propose a multi-scenario instance embedding learning (MultiEmb) framework that implements exclusive feature-dimension redundant information removal for different instances within a scenario to obtain the optimal individual embeddings. The core of our MultiEmb is to introduce an instance embedding selection network to effectively complete the above challenging tasks, in which a set of feature selection and dimension selection adaptive components are equipped for each scenario, and their combination completes the optimal embedding selection for each instance. Finally, we evaluate MultiEmb through extensive experiments on two public multi-scenario benchmarks and demonstrate its effectiveness, compatibility, transferability, etc.

Cross-domain sequential recommendation (CDSR) is proposed to alleviate the data sparsity issue while capturing users' sequential preferences. However, most existing methods do not explore the item transition patterns across different domains and can also not be applied to a multi-domain scenario.Moreover, previous methods rely on overlapping users as bridges to transfer knowledge, which struggles to capture the complex associations across domains without sufficient overlapping users. In this paper, we introduce item attributes into CDSR, and propose a heterogeneous graph transfer learning method to address these issues.Specifically, we construct a cross-domain heterogeneous graph to allow the association of user, item, and category nodes from different domains,and enhance the flexibility of the model by enabling message propagation between more nodes through edge expansion based on the semantic similarity and co-occurrence probability.In addition, we devise meta-paths from different perspectives for nodes at item, user and category levels to guide information aggregation, which can transfer knowledge across domains and reduce the reliance on the number of overlapping users.We further design attention modules to capture users' dynamic preferences from the item sequences they have interacted with in each domain, and explore the transition patterns within category sequences which reflect users' coarse-grained preferences.Finally, we perform knowledge transfer across different domains, and predict the most likely items that users will interact with in each domain. Extensive empirical studies on three real-world datasets indicate that our HGTL significantly outperforms the state-of-the-art baselines in all cases.

Cross-user federated recommendation (CUFR) is a promising solution for providing personalized services without collecting users’ raw data. However, most previous CUFR works mainly focus on providing accurate and privacy-preserving personalized recommendations, but overlook the fact that users can opt out at any time during the training process. In response, we study an emerging and new problem of efficiently training an unlearned model to forget the data of the clients who leave a federated system. It is challenging to simply apply or slightly modify existing machine unlearning or federated unlearning methods to CUFR because of the unique collaboration effect in recommender systems. Although a recent gradient calibration-based method (i.e., FRU) shows promising in training an unlearned model, there are still some limitations: (i) there is a potential possibility that some clients run out of the storage space, (ii) all the remaining clients need to participate in computing the new gradients, (iii) it masks the uniqueness of the local gradients, and (iv) the errors of the calibrated gradients will increase gradually with more iterations. In this article, we propose a novel CUFR unlearning (CUFRU) method. Specifically, we design a gradient transfer station (GTS) module for storing the historical gradients while enabling clients to dynamically participate in the computation of the calibrated gradients with the new gradients based on their online status. Moreover, we design a novel iteration-aware gradient calibration mechanism to strike a balance between the weights of the historical and new gradients at the different stages of the unlearning process, alleviating the calibration errors. Finally, we conduct extensive experiments on three real-world datasets to show that our CUFRU can more efficiently train an unlearned model with the competitive recommendation performance.

Cross-domain recommendation (CDR) aims to alleviate the data sparsity problem by leveraging the benefits of modeling two domains. However, existing research often focuses on the recommendation performance while ignores the privacy leakage issue. We find that an attacker can infer user attribute information from the knowledge (e.g., user preferences) transferred between the source and target domains. For example, in our experiments, the average inference accuracies of attack models on gender and age attributes are 0.8323 and 0.3897. The best-performing attack model achieves accuracies of 0.8847 and 0.4634, exceeding a random inference by 25.10% and 64.04%. We can see that the leakage of user attribute information may significantly exceed what would be expected from random inference. In this paper, we propose a novel recommendation framework named CVGAE (short for camouflaged variational graph autoencoder), which effectively models user behaviors and mitigates the risk of user attribute information leakage at the same time. Specifically, our CVGAE combines the strengths of VAEs in capturing latent features and variability with the ability of GCNs in exploiting high-order relational information. Moreover, to ensure against attribute inference attacks without sacrificing the recommendation performance, we design a user attribute protection module that fuses user attribute-camouflaged information with knowledge transfer during cross-domain processes. We then conduct extensive experiments on three real-world datasets, and find our CVGAE is able to achieve strong privacy protection while making little sacrifices in recommendation accuracy.

Most federated learning-based recommender systems allow clients to access a well-trained high-quality model locally, which provides adversaries with the opportunity to infringe the legitimate copyright of the model. In response, we study an emerging and important problem, i.e., copyright protection of a federated recommendation model, which has not yet been addressed in the community of federated learning or recommender systems. We propose the first backdoor-based ownership verification scheme for federated recommendation (OVFR), which enables the server to claim its ownership for a given suspicious recommendation model. First, we propose to generate a trigger set tailored to recommendation scenarios. In particular, we generate some fake users and items, and then construct a set of fake users with fake interaction records as a trigger set. Moreover, we ensure that the distribution of the popularity of the fake items follows a long-tailed distribution for the effectiveness of the incorporated watermarking. To provide robustness assurance, we propose two different hybrid strategies to make the embeddings of the fake items similar to those of the real items. Second, we focus on effectively learning from a trigger set for recommendation scenarios. In particular, we design an MSE loss function and a contrastive loss function for incorporating the backdoor-based watermarking into the item embeddings, since the item embeddings are often more valuable and easier to be accessed than other parameters of a federated recommendation model. We then design a contrastive loss function to reduce the risk of the fake items being detected. Extensive experiments on three public datasets show the effectiveness of our OVFR in terms of ownership verification, model performance, and robustness.

Different from the data sparsity that traditional recommendations suffer from, context-aware recommender systems (CARS) face specific sparsity challenges related to contextual features, i.e., feature sparsity and interaction sparsity. How knowledge graphs address these challenges remains under-discussed. To bridge this gap, in this article, we first propose a novel pairwise intent graph containing nodes of users, items, entities, and enhanced intents to integrate knowledge graphs into CARS efficiently. Enhanced intent nodes are generated through the specific fusion of relational sub-intent and contextual sub-intent, and they are derived from semantic information and contextual information, respectively. We develop a pairwise intent graph embedding learning (PING) framework based on it. Specifically, our PING uses a pairwise intent joint graph convolution module to obtain refined embedding of all the features, where each enhanced intent node acts as a hub to effectively propagate information among different features and between all the features and knowledge graphs. Then, a recommendation module with refined embeddings is used to replace the randomly initialized embeddings of downstream recommendation models to improve model performance. Extensive experiments on three public datasets and some real-world scenarios verify the effectiveness and compatibility of our PING.

Cross-domain sequential recommendation aims to alleviate the sparsity problem while capturing users’ sequential preferences. However, most existing methods learn the user preferences in each domain separately, and then perform knowledge transfer between them to associate two separated domains, which neglects the item transition patterns across sequences from different domains. Moreover, the sparsity problem still exists since some items in the target and source domains are interacted with only a limited number of times. To address these issues, in this paper we propose a generic framework named multi-view graph contrastive learning (MGCL). Specifically, we tackle the problem from the perspective of an intra-domain item representation view and an inter-domain user preference view. From the former view, we adopt the contrastive mechanism to jointly learn the dynamic sequential information in a user sequence graph and the static collaborative information in the cross-domain global graph, while the latter view is to capture the complementary information of the user’s preferences from different domains. Considering that there are multiple domains in real-world scenarios, we further extend MGCL to MGCL+ for multi-domain sequential recommendation and design multi-domain adaptive gated networks to alleviate the negative transfer problem. Extensive empirical studies on three real-world datasets demonstrate that our MGCL and MGCL+ significantly outperforms the state-of-the-art methods.

Sequential recommendation (SR) and multi-behavior sequential recommendation (MBSR) both come from real-world scenarios. Compared with SR, MBSR takes into account the dependencies of different behaviors. We find that most existing works on MBSR are studied in the context of e-commerce scenarios. In terms of the data format of the behavior types, we observe that the conventional label-formatted data carries limited information and is inadequate for scenarios like social media. With this observation, we introducebehavior set and extend MBSR to behavior set-informed sequential recommendation (BSSR). In BSSR, behavior dependencies become more complex and personalized, and user interest arousal may lack explicit contextual associations. To delve into the dynamics inhered within a behavior set and adaptively tailor recommendation lists upon its variability, we propose a novel solution called Explicit and Implicit modeling via Dual-Path Transformer (EIDP) for BSSR. Our EIDP adopts a dual-path architecture, distinguishing between explicit modeling path (EMP) and implicit modeling path (IMP) based on whether to directly incorporate the behavior representations. EMP features the personalized behavior set-wise transition pattern extractor (PBS-TPE) as its core component. It couples behavioral representations with both the items and positions to explore intra-behavior dynamics within a behavior set at a fine granularity. IMP utilizes light multi-head self-attention blocks (L-MSAB) as encoders under specific behavior types. The obtained multi-view representations are then aggregated by cross-behavior attention fusion (CBAF), using the behavior set of the next time step as a guidance to extract collaborative semantics at the behavioral level. Extensive experiments on two real-world datasets demonstrate the effectiveness of our EIDP. We release the implementation code at: https://github.com/OshiNoCSMA/EIDP.

Sequential recommendation has been widely used to predict users’ potential preferences by learning their dynamic user interests, for which most previous methods focus on capturing item-level dependencies. Despite the great success, they often overlook the stage-level interest dependencies. In real-world scenarios, user interests tend to be staged, e.g., following an item purchase, a user’s interests may undergo a transition into the subsequent phase. And there are intricate dependencies across different stages. Meanwhile, users’ behaviors are usually heterogeneous, including auxiliary behaviors (e.g., examinations) and target behaviors (e.g., purchases), which imply more fine-grained user interests. However, existing methods have limitations in explicitly modeling the relationships between the different types of behaviors. To address the above issues, we propose a novel framework, i.e., dynamic stage-aware user interest learning (DSUIL), for heterogeneous sequential recommendation, which is the first solution to model user interests in a cross-stage manner. Specifically, our DSUIL consists of four modules: (1) a dynamic graph construction module transforms a heterogeneous sequence into several subgraphs to model user interests in a stage-wise manner; (2) a dynamic graph convolution module dynamically learns item representations in each subgraph; (3) a behavior-aware subgraph representation learning module learns the heterogeneous dependencies between behaviors and aggregates item representations to represent the staged user interests; and (4) an interest evolving pattern extractor learns the users’ overall interests for the item prediction. Extensive experimental results on two public datasets show that our DSUIL performs significantly better than the state-of-the-art methods.

Multi-behavior recommender systems (MBRS) have been commonly deployed on real-world industrial platforms for their superior advantages in understanding user preferences and mitigating data sparsity. However, the cascade graph modeling paradigm adopted in mainstream MBRS usually assumes that users will refer to all types of behavioral knowledge they have when making decisions about target behaviors, i.e., use all types of behavioral interactions indiscriminately when modeling and predicting target behaviors for each user. We call this a full decision chain constraint and argue that it may be too strict by ignoring that different types of behavioral knowledge have varying importance for different users. In this paper, we propose a novel automated decision chain selection (AutoDCS) framework to relax this constraint, which can consider each user's unique decision dependencies and select a reasonable set of behavioral knowledge to activate for the prediction of target behavior. Specifically, AutoDCS first integrates some existing MBRS methods in a base cascade module to obtain a set of behavior-aware embeddings. Then, a bilateral matching gating mechanism is used to select an exclusive set of behaviors for the current user-item pair to form a decision chain, and the corresponding behavior-augmented embeddings are selectively activated. Subsequently, AutoDCS combines the behavior-augmented and original behavior-aware embeddings to predict the target behavior. Finally, we evaluate AutoDCS and demonstrate its effectiveness through experiments over four public multi-behavior benchmarks.

Side-information integrated sequential recommendation incorporates supplementary information to alleviate the issue of data sparsity. The state-of-the-art works mainly leverage some side information to improve the attention calculation to learn user representation more accurately. However, there are still some limitations to be addressed in this topic. Most of them merely learn the user representation at the item level and overlook the association of the item sequence and the side-information sequences when calculating the attentions, which results in the incomprehensive learning of user representation. Some of them learn the user representations at both the item and side-information levels, but they still face the problem of insufficient optimization of multiple user representations. To address these limitations, we propose a novel model, i.e., Multi-Sequence Sequential Recommender (MSSR), which learns the user's multiple representations from diverse sequences. Specifically, we design a multi-sequence integrated attention layer to learn more attentive pairs than the existing works and adaptively fuse these pairs to learn user representation. Moreover, our user representation alignment module constructs the self-supervised signals to optimize the representations. Subsequently, they are further refined by our side information predictor during training. For item prediction, our MSSR extra considers the side information of the candidate item, enabling a comprehensive measurement of the user's preferences. Extensive experiments on four public datasets show that our MSSR outperforms eleven state-of-the-art baselines. Visualization and case study also demonstrate the rationality and interpretability of our MSSR.

Multi-scenario recommender systems (MSRSs) have been increasingly used in real-world industrial platforms for their excellent advantages in mitigating data sparsity and reducing maintenance costs. However, conventional MSRSs usually use all relevant features indiscriminately and ignore that different kinds of features have varying importance under different scenarios, which may cause confusion and performance degradation. In addition, existing feature selection methods for deep recommender systems may lack the exploration of scenario relations. In this paper, we propose a novel automated multi-scenario feature selection (MultiFS) framework to bridge this gap, which is able to consider scenario relations and utilize a hierarchical gating mechanism to select features for each scenario. Specifically, MultiFS first efficiently obtains feature importance across all the scenarios through a scenario-shared gate. Then, some scenario-specific gate aims to identify feature importance to individual scenarios from a subset of the former with lower importance. Subsequently, MultiFS imposes constraints on the two gates to make the learning mechanism more feasible and combines the two to select exclusive features for different scenarios. We evaluate MultiFS and demonstrate its ability to enhance the multi-scenario model performance through experiments over two public multi-scenario benchmarks.

Recommendation models are deployed in a variety of commercial applications to provide personalized services for users. However, most of them rely on the users’ original rating records that are often collected by a centralized server for model training, which may cause privacy issues. Recently, some centralized federated recommendation models are proposed for the protection of users’ privacy, which however requires a server for coordination in the whole process of model training. As a response, we propose a novel privacy-aware decentralized federated recommendation (DFedRec) model, which is lossless compared with the traditional model in recommendation performance and is thus more accurate than other models in this line. Specifically, we design a privacy-aware structured client-level graph for the sharing of the model parameters in the process of model training, which is a one-stone-two-bird strategy, i.e., it protects users’ privacy via some randomly sampled fake entries and reduces the communication cost by sharing the model parameters only with the related neighboring users. With the help of the privacy-aware structured client-level graph, we propose two novel collaborative training mechanisms in the setting without a server, including a batch algorithm DFedRec(b) and a stochastic one DFedRec(s), where the former requires the anonymity mechanism while the latter does not. They are both equivalent to probabilistic matrix factorization trained in a centralized server and are thus lossless. We then provide formal analysis of privacy guarantee of our methods and conduct extensive empirical studies on three public datasets with explicit feedback, which show the effectiveness of our DFedRec, i.e., it is privacy aware, communication efficient, and lossless.

Static analysis tools for capturing bugs and vulnerabilities in software programs are widely employed in practice, as they have the unique advantages of high coverage and independence from the execution environment. However, existing tools for analyzing large codebases often produce a great deal of false warnings over genuine bug reports. As a result, developers are required to manually inspect and confirm each warning, a challenging, time-consuming, and automation-essential task. This article advocates a fast, general, and easily extensible approach called Llm4sa that automatically inspects a sheer volume of static warnings by harnessing (some of) the powers of Large Language Models (LLMs). Our key insight is that LLMs have advanced program understanding capabilities, enabling them to effectively act as human experts in conducting manual inspections on bug warnings with their relevant code snippets. In this spirit, we propose a static analysis to effectively extract the relevant code snippets via program dependence traversal guided by the bug warning reports themselves. Then, by formulating customized questions that are enriched with domain knowledge and representative cases to query LLMs, Llm4sa can remove a great deal of false warnings and facilitate bug discovery significantly. Our experiments demonstrate that Llm4sa is practical in automatically inspecting thousands of static warnings from Juliet benchmark programs and 11 real-world C/C++ projects, showcasing a high precision (81.13%) and a recall rate (94.64%) for a total of 9,547 bug warnings. Our research introduces new opportunities and methodologies for using the LLMs to reduce human labor costs, improve the precision of static analyzers, and ensure software trustworthiness

A recommendation system is of vital importance in delivering personalization services, which often brings continuous dual improvement in user experience and organization revenue. However, the data of one single organization may not be enough to build an accurate recommendation model for inactive or new cold-start users. Moreover, due to the recent regulatory restrictions on user privacy and data security, as well as the commercial conflicts, the raw data in different organizations cannot be merged to alleviate the scarcity issue in training a model. In order to learn users’ preferences from such cross-silo data of different organizations and then provide recommendations to the cold-start users, we propose a novel federated learning framework, i.e., federated cross-organization recommendation ecosystem (FedCORE). Specifically, we first focus on the ecosystem problem of cross-organization federated recommendation, including cooperation patterns and privacy protection. For the former, we propose a privacy-aware collaborative training and inference algorithm. For the latter, we define four levels of privacy leakage and propose some methods for protecting the privacy. We then conduct extensive experiments on three real-world datasets and two seminal recommendation models to study the impact of cooperation in our proposed ecosystem and the effectiveness of privacy protection.

Recently, federated recommendation has become a research hotspot mainly because of users’ awareness of privacy in data. As a recent and important recommendation problem, in heterogeneous one-class collaborative filtering (HOCCF), each user may involve of two different types of implicit feedback, that is, examinations and purchases. So far, privacy-preserving HOCCF has received relatively little attention. Existing federated recommendation works often overlook the fact that some privacy sensitive behaviors such as purchases should be collected to ensure the basic business imperatives in e-commerce for example. Hence, the user privacy constraints can and should be relaxed while deploying a recommendation system in real scenarios. In this article, we study the federated multi-behavior recommendation problem under the assumption that purchase behaviors can be collected. Moreover, there are two additional challenges that need to be addressed when deploying federated recommendation. One is the low storage capacity for users’ devices to store all the item vectors, and the other is the low computational power for users to participate in federated learning. To release the potential of privacy-preserving HOCCF, we propose a novel framework, named discrete federated multi-behavior recommendation (DFMR), which allows the collection of the business necessary behaviors (i.e., purchases) by the server. As to reduce the storage overhead, we use discrete hashing techniques, which can compress the parameters down to 1.56% of the real-valued parameters. To further improve the computation-efficiency, we design a memorization strategy in the cache updating module to accelerate the training process. Extensive experiments on four public datasets show the superiority of our DFMR in terms of both accuracy and efficiency.

In recent years, sequential recommendation (SR) has gained significant attention in the recommender systems community. However, most previous works only consider the (user, item, timestep) interaction sequences, which limits the recommendation performance. To overcome this limitation, some studies have utilized user reviews to enrich the understanding of user preferences. However, existing review-based sequential recommendation (RBSR) methods only use either a user’s review on items or an item’s reviews by users, overlooking their complementary nature. In addition, most existing RBSR methods use a simple dot-product operation between the embeddings of a user and the candidate items for scoring, which may not adequately capture the complex relationships among the item sequence, review sequence and candidate items. To release the potential of RBSR, we propose a novel model called cascaded cross attention (CCA), which utilizes aggregated reviews to compensate for the information that is lacking in individual reviews. Moreover, we propose a cascaded cross-attention layer to better capture the dependency intra a sequence and the relationships between a sequence and the candidate items. Extensive experimental results on three public datasets demonstrate that our CCA outperforms the state-of-the-art methods. Additionally, the case study and visualization results showcase high interpretability of our CCA.

Cross-domain sequential recommendation is an important development direction of recommender systems. It combines the characteristics of sequential recommender systems and cross-domain recommender systems, which can capture the dynamic preferences of users and alleviate the problem of cold-start users. However, in recent years, people pay more and more attention to their privacy. How to protect the users’ privacy has become an urgent problem to be solved. In this paper, we propose a novel privacy-preserving cross-domain sequential recommender system (PriCDSR), which can provide users with recommendation services while preserving their privacy at the same time. Specifically, we define a new differential privacy on the data, taking into account both the ID information and the order information. Then, we design a random mechanism that satisfies this differential privacy and provide its theoretical proof. Our PriCDSR is a non-invasive method that can adopt any cross-domain sequential recommender system as a base model without any modification to it. To the best of our knowledge, our PriCDSR is the first work to investigate privacy issues in cross-domain sequential recommender systems. We conduct experiments on three domains, and the results demonstrate that our PriCDSR, despite introducing noise, still outperforms recommender systems that only use data from a single domain.

Variational autoencoder (VAE) is known as a classic and effective method in modeling users’ homogeneous behaviors in recommender systems. In recent years, graph neural networks (GNNs) have achieved promising performance in learning users’ preferences by modeling complex relationships between users and items. However, most VAE- and GNN-based methods are for single-behavior recommendation, rather than the more prevalent counterpart in real-world applications, i.e., multi-behavior recommendation. This motivates us to leverage VAE and GNNs to address the more important and challenging problem of multi-behavior recommendation. Traditional multi-behavior recommendation models have not captured the complex transition relationships across different types of behaviors well. Moreover, they often ignore the varying semantic strength of different types of behaviors. To tackle the above two challenges, we propose a novel multi-behavior recommendation framework named VCGAE (short for Variational Collective Graph AutoEncoder), which inherits the advantages of VAE and GNNs to fully exploit the multi-behavior data. Specifically, we design a behavior transition network that considers different transition relationships across different types of behaviors in order to learn users’ personal preferences. In addition, we introduce a behavior contrastive regularization module that extracts different correlations between users’ auxiliary behaviors and target behavior. By doing so, we gain a deeper understanding of the semantic strength of different behaviors, thereby improving the overall recommendation performance. Extensive experiments on three real-world datasets clearly demonstrate the effectiveness of our VCGAE compared with the state-of-the-art methods.

Although knowledge graph has shown their effectiveness in mitigating data sparsity in many recommendation tasks, they remain underutilized in context-aware recommender systems (CARS) with the specific sparsity challenges associated with the contextual features, i.e., feature sparsity and interaction sparsity. To bridge this gap, in this paper, we propose a novel pairwise intent graph embedding learning (PING) framework to efficiently integrate knowledge graphs into CARS. Specifically, our PING contains three modules: 1) a graph construction module is used to obtain a pairwise intent graph (PIG) containing nodes for users, items, entities, and enhanced intent, where enhanced intent nodes are generated by applying user intent fusion (UIF) on relational intent and contextual intent, and two sub-intents are derived from the semantic information and contextual information, respectively; 2) a pairwise intent joint graph convolution module is used to obtain the refined embeddings of all the features by executing a customized convolution strategy on PIG, where each enhanced intent node acts as a hub to efficiently propagate information among different features and between all the features and knowledge graph; 3) a recommendation module with the refined embeddings is used to replace the randomly initialized embeddings of downstream recommendation models to improve model performance. Finally, we conduct extensive experiments on three public datasets to verify the effectiveness and compatibility of our PING.

A practical recommender system should be able to handle heterogeneous behavioral feedback as inputs and has multi-task outputs ability. Although the heterogeneous one-class collaborative filtering (HOCCF) and multi-task learning (MTL) methods has been well studied, there is still a lack of targeted manner in their combined fields, i.e., Multi-behavior Multi-task Recommendation (MMR). To fill the gap, we propose a novel recommendation framework called Behavior-aware Variational AutoEncoder (BVAE), which meliorates the parameter sharing and loss minimization method with the VAE structure to address the MMR problem. Specifically, our BVAE includes behavior-aware semi-encoders and decoders, and a target feature fusion network with a global feature filtering network, while using standard deviation to weigh loss. These modules generate the behavior-aware recommended item list via constructing better semantic feature vectors for users, i.e., from dual perspectives of behavioral preference and global interaction. In addition, we optimize our BVAE in terms of adaptability and robustness, i.e., it is concise and flexible to consume any amount of behaviors with different distributions. Extensive empirical studies on two real and widely used datasets confirm the validity of our design and show that our BVAE can outperform the state-of-the-art related baseline methods under multiple evaluation metrics. The processed datasets, source code, and scripts necessary to reproduce the results can be available at https://github.com/WitnessForest/BVAE.

Sequential recommendation methods play an irreplaceable role in recommender systems which can capture the users’ dynamic preferences from the behavior sequences. Despite their success, these works usually suffer from the sparsity problem commonly existed in real applications. Cross-domain sequential recommendation aims to alleviate this problem by introducing relatively richer source-domain data. However, most existing methods capture the users’ preferences independently of each domain, which may neglect the item transition patterns across sequences from different domains, i.e., a user’s interaction in one domain may influence his/her next interaction in other domains. Moreover, the data sparsity problem still exists since some items in the target and source domains are interacted with only a limited number of times. To address these issues, in this paper we propose a generic framework named multi-view graph contrastive learning (MGCL). Specifically, we adopt the contrastive mechanism in an intra-domain item representation view and an inter-domain user preference view. The former is to jointly learn the dynamic sequential information in the user sequence graph and the static collaborative information in the cross-domain global graph, while the latter is to capture the complementary information of the user’s preferences from different domains. Extensive empirical studies on three real-world datasets demonstrate that our MGCL significantly outperforms the state-of-the-art methods.

Tabular data is one of the most common data storage formats behind many real-world web applications such as retail, banking, and e-commerce. The success of these web applications largely depends on the ability of the employed machine learning model to accurately distinguish influential features from all the predetermined features in tabular data. Intuitively, in practical business scenarios, different instances should correspond to different sets of influential features, and the set of influential features of the same instance may vary in different scenarios. However, most existing methods focus on global feature selection assuming that all instances have the same set of influential features, and few methods considering instance-wise feature selection ignore the variability of influential features in different scenarios. In this paper, we first introduce a new perspective based on the influence function for instance-wise feature selection, and give some corresponding theoretical insights, the core of which is to use the influence function as an indicator to measure the importance of an instance-wise feature. We then propose a new solution for discovering instance-wise influential features in tabular data (DIWIFT), where a self-attention network is used as a feature selection model and the value of the corresponding influence function is used as an optimization objective to guide the model. Benefiting from the advantage of the influence function, i.e., its computation does not depend on a specific architecture and can also take into account the data distribution in different scenarios, our DIWIFT has better flexibility and robustness. Finally, we conduct extensive experiments on both synthetic and real-world datasets to validate the effectiveness of our DIWIFT.

Heterogeneous sequential recommendation that models sequences of items associated with more than one type of feedback such as examinations and purchases is an emerging topic in the research community, which is also an important problem in many real-world applications. Though there are some methods proposed to exploit different types of feedback in item sequences such as RLBL, RIB, and BINN, they are based on RNN and may not be very competitive in capturing users’ complex and dynamic preferences. And most existing advanced sequential recommendation methods such as the CNN- and attention-based methods are often designed for making use of item sequences with one single type of feedback, which thus can not be applied to the studied problem directly. As a response, we propose a novel feedback-aware local and global (FLAG) preference learning model for heterogeneous sequential recommendation. Our FLAG contains four modules, including (i) a local preference learning module for capturing a user’s short-term interest, which adopts a novel feedback-aware self-attention block to distinguish different types of feedback; (ii) a global preference learning module for modeling a user’s global preference; (iii) a local intention learning module, which takes a user’s real feedback in the next step, i.e., the user’s intention at the current step, as the query vector in a self-attention block to figure out the items that match the user’s intention well; and (iv) a prediction module for preference integration and final prediction. We then conduct extensive experiments on three public datasets and find that our FLAG significantly outperforms 13 very competitive baselines in terms of two commonly used ranking-oriented metrics in most cases. We also include ablation studies and sensitivity analysis of our FLAG to have more in-depth insights.

The bias problems in recommender systems are an important challenge. In this paper, we focus on solving the bias problems via uniform data. Previous works have shown that simple modeling with a uniform data can alleviate the bias problems and improve the performance. However, the uniform data is usually few and expensive to collect in a real product. In order to use the valuable uniform data more effectively, we propose a novel and general knowledge distillation framework for counterfactual recommendation with four specific methods, including label-based distillation, feature-based distillation, sample-based distillation and model structure-based distillation. Moreover, we discuss the relation between the proposed framework and the previous works. We then conduct extensive experiments on both public and product datasets to verify the effectiveness of the proposed four methods. In addition, we explore and analyze the performance trends of the proposed methods on some key factors, and the changes in the distribution of the recommendation lists. Finally, we emphasize that counterfactual modeling with uniform data is a rich research area, and list some interesting and promising research topics worthy of further exploration. Note that the source codes are available athttps://github.com/dgliu/TKDE_KDCRec.

With the implementation of privacy protection laws such as GDPR, it is increasingly difficult for organizations to legally collect users’ data. However, a typical machine learning-based recommendation algorithm requires the data to learn users’ preferences. Some recent works thus turn to develop federated learning-based recommendation algorithms, but most of them either cannot protect the users’ privacy well, or sacrifice the model accuracy. In this article, we propose a lossless and generic federated recommendation framework via fake marks and secret sharing (FMSS). Our FMSS can not only protect the two types of users’ privacy, i.e., rating values and rating behaviors, without sacrificing the recommendation performance, but can also be applied to most recommendation algorithms for rating prediction, item ranking, and sequential recommendation. Specifically, we extend existing fake items to fake marks, and combine it with secret sharing to perturb the data uploaded by the clients to a server. We then apply our FMSS to six representative recommendation algorithms, i.e., MF-MPC and NeuMF for rating prediction, eALS and VAE-CF for item ranking, and Fossil and GRU4Rec for sequential recommendation. The experimental results demonstrate that our FMSS is a lossless and generic framework, which is able to federate a series of different recommendation algorithms in a lossless and privacy-aware manner.

Debiased recommendation with a randomized dataset has shown very promising results in mitigating system-induced biases. However, it still lacks more theoretical insights or an ideal optimization objective function compared with the other more well-studied routes without a randomized dataset. To bridge this gap, we study the debiasing problem from a new perspective and propose to directly minimize the upper bound of an ideal objective function, which facilitates a better potential solution to system-induced biases. First, we formulate a new ideal optimization objective function with a randomized dataset. Second, according to the prior constraints that an adopted loss function may satisfy, we derive two different upper bounds of the objective function: a generalization error bound with triangle inequality and a generalization error bound with separability. Third, we show that most existing related methods can be regarded as the insufficient optimization of these two upper bounds. Fourth, we propose a novel method called debiasing approximate upper bound ( DUB ) with a randomized dataset, which achieves a more sufficient optimization of these upper bounds. Finally, we conduct extensive experiments on a public dataset and a real product dataset to verify the effectiveness of our DUB.

How to effectively mitigate the bias of feedback in recommender systems is an important research topic. In this article, we first describe the generation process of the biased and unbiased feedback in recommender systems via two respective causal diagrams, where the difference between them can be regarded as the source of system-induced biases. We then define this difference as a confounding bias and propose a new perspective on debiased representation learning to alleviate it. Specifically, for the case with biased feedback alone, we derive the conditions that need to be satisfied to obtain a debiased representation from the causal diagrams. Then, we propose a novel framework called debiased information bottleneck (DIB) to optimize these conditions and then find a tractable solution for it. The proposed framework constrains the model to learn a biased embedding vector with independent biased and unbiased components in the training phase, and uses only the unbiased component in the test phase to deliver more accurate recommendations. We further propose a variant of DIB by relaxing the independence between the biased and unbiased components. Finally, we conduct extensive experiments on a public dataset and a real product dataset to verify the effectiveness of the proposed framework.

Recently, sequential recommendation has become a research hotspot while multi-behavior sequential recommendation (MBSR) that exploits users' heterogeneous interactions in sequences has received relatively little attention. Existing works often overlook the complementary effect of different perspectives when addressing the MBSR problem. In addition, there are two specific challenges remained to be addressed. One is the heterogeneity of a user's intention and the context information, the other one is the sparsity of the interactions of target behavior. To release the potential of multi-behavior interaction sequences, we propose a novel framework named NextIP that adopts a dual-task learning strategy to convert the problem to two specific tasks, i.e., next-item prediction and purchase prediction. For next-item prediction, we design a target-behavior aware context aggregator (TBCG), which utilizes the next behavior to guide all kinds of behavior-specific item sub-sequences to jointly predict the next item. For purchase prediction, we design a behavior-aware self-attention (BSA) mechanism to extract a user's behavior-specific interests and treat them as negative samples to learn the user's purchase preferences. Extensive experimental results on two public datasets show that our NextIP performs significantly better than the state-of-the-art methods.

Autoencoder-based methods have achieved significant performance on item recommendation. However, they may not perform well on tail items due to the ignorance of the items’ popularity bias. As a response, in this paper, we focus on tail items and propose a novel adversarial learning method for tail recommendation (ALTRec). In our ALTRec, the generator (i.e., AutoRec) not only reconstructs the input well, but also minimizes the (any two-user) similarity difference between the input stage and the output stage to keep users’ interaction relationships unchanged. And the discriminator maps the inputs and outputs of the generator to a same semantic space for scoring the similarity and maximizes the similarity difference as the target, and will identify some unsatisfactory predictions, especially on tail items. In order to preserve the similarity, the generator will pay more attention to the tail items compared with the previous autoencoder-based methods. An ablation study validates the effectiveness of preserving the two-user similarity, as well as the adversarial learning strategy in our ALTRec. Extensive experiments on three real-world datasets show that our ALTRec significantly boosts the performance on tail items compared with several state-of-the-art methods.

Most methods for context-aware recommendation focus on improving the feature interaction layer, but overlook the embedding layer. However, an embedding layer with random initialization often suffers in practice from the sparsity of the contextual features, as well as the interactions between the users (or items) and context. In this paper, we propose a novel user-event graph embedding learning (UEG-EL) framework to address these two sparsity challenges. Specifically, our UEG-EL contains three modules: 1) a graph construction module is used to obtain a user-event graph containing nodes for users, intents and items, where the intent nodes are generated by applying intent node attention (INA) on nodes of the contextual features; 2) a user-event collaborative graph convolution module is designed to obtain the refined embeddings of all features by executing a new convolution strategy on the user-event graph, where each intent node acts as a hub to efficiently propagate the information among different features; 3) a recommendation module is equipped to integrate some existing context-aware recommendation model, where the feature embeddings are directly initialized with the obtained refined embeddings. Moreover, we identify a unique challenge of the basic framework, that is, the contextual features associated with too many instances may suffer from noise when aggregating the information. We thus further propose a simple but effective variant, i.e., UEG-EL-V, in order to prune the information propagation of the contextual features. Finally, we conduct extensive experiments on three public datasets to verify the effectiveness and compatibility of our UEG-EL and its variant.

Heterogeneous sequential recommendation (HSR) is a very important recommendation problem, which aims to predict a user’s next interacted item under a target behavior type (e.g., purchase in e-commerce sites) based on his/her historical interactions with different behaviors. Though existing sequential methods have achieved advanced performance by considering the varied impacts of interactions with sequential information, a large body of them still have two major shortcomings. Firstly, they usually model different behaviors separately without considering the correlations between them. The transitions from item to item under diverse behaviors indicate some users’ potential behavior manner. Secondly, though the behavior information contains a user’s fine-grained interests, the insufficient consideration of the local context information limits them from well understanding user intentions. Utilizing the adjacent interactions to better understand a user’s behavior could improve the certainty of prediction. To address these two issues, we propose a novel solution utilizing global and personalized graphs for HSR (GPG4HSR) to learn behavior transitions and user intentions. Specifically, our GPG4HSR consists of two graphs, i.e., a global graph to capture the transitions between different behaviors, and a personalized graph to model items with behaviors by further considering the distinct user intentions of the adjacent contextually relevant nodes. Extensive experiments on four public datasets with the state-of-the-art baselines demonstrate the effectiveness and general applicability of our method GPG4HSR.

Neural network-based models for collaborative filtering have received widespread attention, among which variational autoencoder (VAE) has shown unique advantages in the task of item recommendation. However, most existing VAE-based models only focus on one type of user feedback, leading to their performance bottlenecks. To overcome this limitation, we propose a novel VAE-based recommendation model called VAE++, which can effectively utilize heterogeneous feedback to boost recommendation performance. Specifically, it combines three different types of signals, i.e., purchase feedback, examination feedback and their mixed feedback, via two well-designed modules, i.e., a target representation enhancement module and a target representation refinement module. The former exploits the mixed feedback to improve the learning of purchase representations, while the latter leverages the examination feedback to further refine them. In particular, purchase and examination preferences are jointly decoded in one decoder to ensure the high quality of the reconstructed samples. Extensive experiments on three public datasets show that our VAE++ achieves the best results compared with several state-of-the-art methods.

Modeling other agents' behaviors plays an important role in decision models for interactions among multiple agents. To optimize its own decisions, a subject agent needs to model what other agents act simultaneously in an uncertain environment. However, modeling insufficiency occurs when the agents are competitive and the subject agent cannot get full knowledge about other agents. Even when the agents are collaborative, they may not share their true behaviors due to their privacy concerns. Most of the recent research still assumes that the agents have common knowledge about their environments and a subject agent has the true behavior of other agents in its mind. Consequently, the resulting techniques are not applicable in many practical problem domains. In this article, we investigate into diversifying behaviors of other agents in the subject agent's decision model before their interactions. The challenges lie in generating and measuring new behaviors of other agents. Starting with prior knowledge about other agents' behaviors, we use a linear reduction technique to extract representative behavioral features from the known behaviors. We subsequently generate their new behaviors by expanding the features and propose two diversity measurements to select top- K $K$ behaviors. We demonstrate the performance of the new techniques in two well-studied problem domains. The top- K $K$ behavior selection embarks the study of unknown behaviors in multiagent decision making and inspires investigation of diversifying agents' behaviors in competitive agent interactions. This study will contribute to intelligent systems dealing with unknown unknowns in an open artificial intelligence world.

In a recommender system, a user’s interaction is often biased by the items’ displaying positions and popularity, as well as the user’s self-selection. Most existing recommendation models are built using such a biased user-system interaction data. In this paper, we first additionally introduce a specially collected unbiased data and then propose a novel transfer learning solution, i.e., transfer via joint reconstruction (TJR), to achieve knowledge transfer and sharing between the biased data and unbiased data. Specifically, in our TJR, we refine the prediction via the latent features containing bias information in order to obtain a more accurate and unbiased prediction. Moreover, we integrate the two data by reconstructing their interaction in a joint learning manner. We then adopt three representative methods as the backbone models of our TJR and conduct extensive empirical studies on two public datasets, showcasing the effectiveness of our transfer learning solution over some very competitive baselines.

With the implementation of privacy protection laws such as GDPR, it is increasingly difficult for organizations to legally collect user data. However, a typical recommendation algorithm based on machine learning requires user data to learn user preferences. In order to protect user privacy, a lot of recent works turn to develop federated learning-based recommendation algorithms. However, some of these works can only protect the users’ rating values, some can only protect the users’ rating behavior (i.e., the engaged items), and only a few works can protect the both types of privacy at the same time. Moreover, most of them can only be applied to a specific algorithm or a class of similar algorithms. In this paper, we propose a generic cross-user federated recommendation framework called FR-FMSS. Our FR-FMSS can not only protect the two types of user privacy, but can also be applied to most recommendation algorithms for rating prediction, item ranking, and sequential recommendation. Specifically, we use fake marks and secret sharing to modify the data uploaded by the clients to the server, which protects user privacy without loss of model accuracy. We take three representative recommendation algorithms, i.e., MF-MPC, eALS, and Fossil, as examples to show how to apply our FR-FMSS to a specific algorithm.

How to effectively mitigate the bias of feedback in recommender systems is an important research topic. In this paper, we first describe the generation process of the biased and unbiased feedback in recommender systems via two respective causal diagrams, where the difference between them can be regarded as the source of bias. We then define this difference as a confounding bias, which can be regarded as a collection of some specific biases that have previously been studied. For the case with biased feedback alone, we derive the conditions that need to be satisfied to obtain a debiased representation from the causal diagrams. Based on information theory, we propose a novel method called debiased information bottleneck (DIB) to optimize these conditions and then find a tractable solution for it. In particular, the proposed method constrains the model to learn a biased embedding vector with independent biased and unbiased components in the training phase, and uses only the unbiased component in the test phase to deliver more accurate recommendations. Finally, we conduct extensive experiments on a public dataset and a real product dataset to verify the effectiveness of the proposed method and discuss its properties.

Abstract With the availability of significant amount of data, data-driven decision making becomes an alternative way for solving complex multiagent decision problems. Instead of using domain knowledge to explicitly build decision models, the data-driven approach learns decisions (probably optimal ones) from available data. This removes the knowledge bottleneck in the traditional knowledge-driven decision making, which requires a strong support from domain experts. In this paper, we study data-driven decision making in the context of interactive dynamic influence diagrams (I-DIDs)—a general framework for multiagent sequential decision making under uncertainty. We propose a data-driven framework to solve the I-DIDs model and focus on learning the behavior of other agents in problem domains. The challenge is on learning a complete policy tree that will be embedded in the I-DIDs models due to limited data. We propose two new methods to develop complete policy trees for the other agents in the I-DIDs. The first method uses a simple clustering process, while the second one employs sophisticated statistical checks. We analyze the proposed algorithms in a theoretical way and experiment them over two problem domains.

Sequential recommendation has been a hot research topic because of its practicability and high accuracy by capturing the sequential information. As deep learning (DL) based methods being widely adopted to model the local and dynamic preferences beneath users’ behavior sequences, the modeling of users’ global and static preferences tends to be underestimated that usually, only some simple and crude users’ latent representations are introduced. Moreover, most existing methods hold an assumption that users’ intention can be fully captured by considering the historical behaviors, while neglect the possible uncertainty of users’ intention in reality, which may be influenced by the appearance of the candidate items to be recommended. In this paper, we thus focus on these two issues, i.e., the imperfect modeling of users’ global preferences in most DL-based sequential recommendation methods and the uncertainty of users’ intention brought by the candidate items, and propose a novel solution named fusing item similarity models with self-attention networks (FISSA) for sequential recommendation. Specifically, we treat the state-of-the-art self-attentive sequential recommendation (SASRec) model as the local representation learning module to capture the dynamic preferences beneath users’ behavior sequences in our FISSA, and further propose a global representation learning module to improve the modeling of users’ global preferences and a gating module that balances the local and global representations by taking the information of the candidate items into account. The global representation learning module can be seen as a location-based attention layer, which is effective to fit in well with the parallelization training process of the self-attention framework. The gating module calculates the weight by modeling the relationship among the candidate item, the recently interacted item and the global preference of each user using an MLP layer. Extensive empirical studies on five commonly used datasets show that our FISSA significantly outperforms eight state-of-the-art baselines in terms of two commonly used metrics.

Recommender systems are feedback loop systems, which often face bias problems such as popularity bias, previous model bias and position bias. In this paper, we focus on solving the bias problems in a recommender system via a uniform data. Through empirical studies in online and offline settings, we observe that simple modeling with a uniform data can alleviate the bias problems and improve the performance. However, the uniform data is always few and expensive to collect in a real product. In order to use the valuable uniform data more effectively, we propose a general knowledge distillation framework for counterfactual recommendation that enables uniform data modeling through four approaches: (1) label-based distillation focuses on using the imputed labels as a carrier to provide useful de-biasing guidance; (2) feature-based distillation aims to filter out the representative causal and stable features; (3) sample-based distillation considers mutual learning and alignment of the information of the uniform and non-uniform data; and (4) model structure-based distillation constrains the training of the models from the perspective of embedded representation. We conduct extensive experiments on both public and product datasets, demonstrating that the proposed four methods achieve better performance over the baseline models in terms of AUC and NLL. Moreover, we discuss the relation between the proposed methods and the previous works. We emphasize that counterfactual modeling with uniform data is a rich research area, and list some interesting and promising research topics worthy of further exploration. Note that the source codes are available at \urlhttps://github.com/dgliu/SIGIR20_KDCRec.

With the explosive growth of web resources, an increasingly important task in recommender systems is to provide high-quality personalized services by learning users’ preferences from historically observed information. As an effective preference learning technology, collaborative filtering has been widely extended to model the one-class or implicit feedback data, which is known as one-class collaborative filtering (OCCF). For a long time, pairwise ranking-oriented learning scheme has been viewed as a superior solution than the pointwise scheme for OCCF due to its higher accuracy in most cases. However, we argue that with appropriate model design, pointwise preference learning can achieve comparable or even better performance than the counterpart, i.e., pairwise preference learning. In particular, we propose a new preference assumption, i.e., pointwise preference on user/item-set. Based on this new assumption, we develop a novel, simple, and flexible solution called collaborative filtering via pointwise preference learning on user/item-set (CoFi-points). Furthermore, we derive two specific algorithms of CoFi-points with respect to the involved user-set and item-set, i.e., CoFi-points(u) and CoFi-points(i), referring to preference assumptions defined on user-set and item-set, respectively. Finally, we conduct extensive empirical studies on four real-world datasets with the state-of-the-art methods, and find that our solution can achieve very promising performance with respect to several ranking-oriented evaluation metrics.

With the rapid increase of the data size, it has increasing demands for selecting features by exploiting both labeled and unlabeled data. In this paper, we propose a novel semi-supervised embedded feature selection method. The new method extends the least square regression model by rescaling the regression coefficients in the least square regression with a set of scale factors, which is used for evaluating the importance of features. An iterative algorithm is proposed to optimize the new model. It has been proved that solving the new model is equivalent to solving a sparse model with a flexible and adaptable ℓ2;pnorm regularization. Moreover, the optimal solution of scale factors provides a theoretical explanation for why we can use {||w1||2, . . .,||wd||2} to evaluate the importance of features. Experimental results on eight benchmark data sets show the superior performance of the proposed method.

Recommender systems play an important role in providing personalized services for users in the context of information overload. Generally, users’ feedback toward items often contain the most significant information reflecting their preferences, which enables accurate personalized recommendation. In real applications, users’ feedback are usually heterogeneous (rather than homogeneous) such as purchases and examinations in e-commerce, which reflects users’ preferences in different degrees. Effective modeling of such heterogeneous one-class feedback is challenging compared with that of homogeneous feedback of ratings. As a response, heterogeneous one-class collaborative filtering (HOCCF) is proposed, which often converts the heterogeneous feedback into two parts (i.e., target feedback and auxiliary feedback), aiming to care more about the target feedback (e.g., purchases ) with the assistance of the auxiliary feedback (e.g., examinations ). In this survey, we provide an overview of the representative HOCCF methods from the perspective of factorization-based methods, transfer learning-based methods, and deep learning-based methods. First, we review the factorization-based methods according to different strategies. Second, we describe the transfer learning-based methods with different knowledge sharing manners. Third, we discuss the deep learning-based methods according to the neural architectures. Moreover, we include some important example applications, describe the empirical studies, and discuss some promising future directions.

Exploiting temporal effect has empirically been recognized as a promising way to improve recommendation performance in recent years. In real-world applications, one-class data in the form of (user, item, timestamp) are usually more accessible and abundant than numerical ratings. In this article, we focus on exploiting such one-class data in order to provide personalized next-item recommendation services. Specifically, we base our work on the framework of time-aware item-based collaborative filtering and propose a simple yet effective similarity measurement called bidirectional item similarity (BIS) that is able to capture sequential patterns even from noisy data. Furthermore, we extend BIS via some factorization techniques and obtain an adaptive version, i.e., adaptive BIS (ABIS), in order to better fit the behavioral data. We also design a compound weighting function that leverages the complementarity between two well-known time-aware weighting functions. With the proposed similarity measurements and weighting function, we obtain two novel collaborative filtering methods that are able to achieve significantly better performance than the state-of-the-art methods, showcasing their effectiveness for next-item recommendation.

Enumerating maximal cliques from an uncertain graph is a fundamental problem in uncertain graph analysis. Given an uncertain graph G, a set of nodes C in G is a maximal (k, τ)-clique if (1) |C|>k and C is a clique with probability at least τ, and (2) C is a maximal node set meeting (1). The state-of-the-art algorithm for enumerating all maximal (k, τ)-cliques is very costly when handling large uncertain graphs, as its time complexity is proportional to 2^n where n is the number of nodes in the uncertain graph. To overcome this issue, we propose two new core-based pruning algorithms to reduce the uncertain graph size without missing any maximal (k, τ)-clique. We also develop a novel cut-based optimization technique to further improve the pruning performance of the core-based pruning algorithms. Based on these pruning techniques, we propose an improved algorithm to enumerate all maximal (k, τ)-cliques, and a new algorithm with several novel upper-bounding techniques to compute one of maximum (k, τ)-cliques from the pruned uncertain graph. The results of extensive experiments on six real-world datasets demonstrate the efficiency and effectiveness of the proposed algorithms.

In this paper, we propose a novel preference assumption for modeling users' one-class feedback such as "thumb up" in an important recommendation problem called one-class collaborative filtering (OCCF). Specifically, we address a fundamental limitation of a recent symmetric pairwise preference assumption and propose a novel and first asymmetric one, which is able to make the preferences of different users more comparable. With the proposed asymmetric pairwise preference assumption, we further design a novel recommendation algorithm called asymmetric Bayesian personalized ranking (ABPR). Extensive empirical studies on two large and public datasets show that our ABPR performs significantly better than several state-of-the-art recommendation methods with either pointwise preference assumption or pairwise preference assumption.

Heterogeneous one-class collaborative filtering is an emerging and important problem in recommender systems, where two different types of one-class feedback, i.e., purchases and browses, are available as input data. The associated challenges include ambiguity of browses, scarcity of purchases, and heterogeneity arising from different feedback. In this article, we propose to model purchases and browses from a new perspective, i.e., users’ roles of mixer, browser and purchaser. Specifically, we design a novel transfer learning solution termed role-based transfer to rank (RoToR), which contains two variants, i.e., integrative RoToR and sequential RoToR. In integrative RoToR, we leverage browses into the preference learning task of purchases, in which we take each user as a sophisticated customer (i.e., mixer ) that is able to take different types of feedback into consideration. In sequential RoToR, we aim to simplify the integrative one by decomposing it into two dependent phases according to a typical shopping process. Furthermore, we instantiate both variants using different preference learning paradigms such as pointwise preference learning and pairwise preference learning. Finally, we conduct extensive empirical studies with various baseline methods on three large public datasets and find that our RoToR can perform significantly more accurate than the state-of-the-art methods.

Intelligent recommendation has been well recognized as one of the major approaches to address the information overload problem in the big data era. A typical intelligent recommendation engine usually consists of three major components, that is, data as the main input, algorithms for preference learning, and system for user interaction and high-performance computation. We observe that the data (e.g., users’ behavior) are usually in different forms, such as examinations (e.g., browse and collection) and ratings, where the former are often much more abundant than the latter. Although the data are in different representations, they are both related to users’ true preferences and are also deemed complementary to each other for preference learning. However, very few ranking or recommendation algorithms have been developed to exploit such two types of user behavior. In this article, we focus on jointly modeling the examination behavior and rating behavior and develop a novel and efficient ranking-oriented recommendation algorithm accordingly. First, we formally define a new recommendation problem termed behavior ranking , which aims to build a ranking-oriented model by exploiting both the examination behavior and rating behavior. Second, we develop a simple and generic transfer to rank (ToR) algorithm for behavior ranking, which transfers knowledge of candidate items from a global preference learning task to a local preference learning task. Compared with the previous work on integrating heterogeneous user behavior, our ToR algorithm is the first ranking-oriented solution, which can effectively generate recommendations in a more direct manner than those regression-oriented methods. Extensive empirical studies show that our ToR algorithm performs significantly more accurately than the state-of-the-art methods in most cases. Furthermore, our ToR algorithm is very efficient in terms of the time complexity, which is similar to those for homogeneous user behavior alone.

Learning users' preferences is critical to enable personalized recommendation services in various online applications such as e-commerce, entertainment and many others. In this paper, we study on how to learn users' preferences from abundant online activities, e.g., browsing and examination, which are usually called implicit feedbacks since they cannot be interpreted as users' likes or dislikes on the corresponding products directly. Pairwise preference learning algorithms are the state-of-the-art methods for this important problem, but they have two major limitations of low accuracy and low efficiency caused by noise in observed feedbacks and non-optimal learning steps in update rules. As a response, we propose a novel adaptive pairwise preference learning algorithm, which addresses the above two limitations in a single algorithm with a concise and general learning scheme. Specifically, in the proposed learning scheme, we design an adaptive utility function and an adaptive learning step for the aforementioned two problems, respectively. Empirical studies show that our algorithm achieves significantly better results than the state-of-the-art method on two real-world data sets.

Users often use tags to annotate and categorize web content. A folksonomy is a system of classification derived from the practice and method of collaboratively creating and managing tags. The most significant feature of a folksonomy is that it directly reflects the vocabulary of users. This feature is very useful in tag-based content searching and user browsing. Based on mutual-overlapping measurement of tag's instance sets, an ontology learning algorithm to construct concept hierarchy from folksonomy is proposed. A case study of datasets from a famous Chinese e-business website taobao is carried out. The precision, valid, recall and F-measure rates of the constructed concept hierarchy are 54%, 84%, 100% and 70% respectively. The experimental results on real world datasets show that the proposed method is feasible.