Graph Contrastive Learning (GCL) has recently garnered significant attention for enhancing recommender systems. Most existing GCL-based methods perturb the raw data graph to generate views, performing contrastive learning across these views to learn generalizable representations. However, most of these methods rely on data- or model-based augmentation techniques that may disrupt interest consistency. In this paper, we propose a novel interest-aware augmentation approach based on diffusion models to address this issue. Specifically, we leverage a conditional diffusion model to generate interest-consistent views by conditioning on node interaction information, ensuring that the generated views align with the interests of the nodes. Based on this augmentation method, we introduce DiffCL, a graph contrastive learning framework for recommendation. Furthermore, we propose an easy-to-hard generation strategy. By progressively adjusting the starting point of the reverse denoising process, this strategy further enhances effective contrastive learning. We evaluate DiffCL on three public real-world datasets, and results indicate that our method outperforms state-of-the-art techniques, demonstrating its effectiveness.

Designing effective negative sampling strategies is crucial for training Multi-Modal Recommendation (MMRec) models, as it helps address the issues of sparse user-item interactions and facilitates the learning of high-dimensional modality features. However, most existing methods randomly sample non-interacted items as negative ones, which frequently result in easy negatives. They limit the model's ability to accurately capture user preferences. In this paper, we propose to Generate Difficulty-aware Negative Samples via conditional diffusion for MMRec (denoted as GDNSM). Leveraging the rich semantic and contextual information from multi-modal features, our method generates hard negative samples with varying difficulty levels, tailored to user preferences. They force the model to learn finer-grained distinctions between positive and negative samples, enhancing its ability to inferring user preferences. To avoid unstable training, we design a dynamic difficulty scheduling mechanism that schedules the negative samples from easy to hard for model training, ensuring both stability and effectiveness. Extensive experiments on three real-world datasets demonstrate that the effectiveness of our models.

User cold-start recommendation presents a significant challenge for recommender systems, affecting their overall effectiveness. Meta-learning-based methods have been introduced to address this issue. These methods treat the user cold-start recommendation problem as a few-shot learning task, where each user represents a unique task. The objective is to acquire shared initialization parameters that can be effectively applied across all cold-start users. Subsequently, these shared parameters are fine-tuned into personalized parameters using individual interaction data. Recent studies argue that shared parameters are unsuitable for all users with an implicit grouping distribution of user preference. Therefore, they propose adaptive-initialization-based methods, which first differentiate tasks based on user preferences and then generate task-adaptive initialization parameters using task representations. However, both the meta-learning and adaptive-initialization-based manners ignore discovering the adaptive capability of update strategies in the process of transferring initialization parameters to personalized parameters. Instead, they rely on task-shared optimization strategies, leading the model to fall into an overfitting or underfitting situation. In response to this, we propose a dual-adaptive update strategies-enhanced meta-optimization framework (DAUS) for user cold-start recommendation. First, we integrate dual-adaptive update strategies to enhance the adaptive capability of transferring initialization parameters. This involves incorporating both task-adaptive optimization hyperparameters and objectives. Second, we design a multifaceted task encoder , which can provide diverse task information to differentiate between tasks, including explicit task features (task relevance, training signals) and other implicit task information. Extensive experiments based on three real-world datasets demonstrate that our DAUS outperforms the state-of-the-art methods. The source code is available at https://github.com/XuHao-bit/DAUS .

Designed to establish potential relations and distill high-order representations, graph-based recommendation systems continue to reveal promising results by jointly modeling ratings and reviews. However, existing studies capture simple review relations, failing to (1) completely explore hidden connections between users (or items), (2) filter out redundant information derived from reviews, and (3) model the behavioral association between rating and review interactions. To address these challenges, we propose a review-enhanced hierarchical contrastive learning, namely ReHCL. First, ReHCL constructs topic and semantic graphs to fully mine review relations from different views. Moreover, a cross-view graph contrastive learning is used to achieve enhancement of node representations and extract useful review knowledge. Meanwhile, we design a neighbor-based positive sampling to capture the graph-structured similarity between topic and semantic views, further performing efficient contrast and reducing redundant noise. Next, we propose a cross-modal contrastive learning to match the rating and review representations, by exploring the association between ratings and reviews. Lastly, these two contrastive learning modes form a hierarchical contrastive learning task, which is applied to enhance the final recommendation task. Extensive experiments verify the superiority of ReHCL compared with state-of-the-arts.

POI recommendation aims to learn diverse user preferences based on their historical behavioral trajectories and to recommend locations that align with user interests. In this scenario, discerning behavioral patterns and preferences from user check-in sequences plays a crucial role. In recent study, Markov Chain-based methods mainly focus on modeling transition relation between items in a sequence, RNN-based and attention-based methods pay great attention to capture sequence patterns and user’s preferences. However, existing methods neglect the significance of the user’s current intent in next POI recommendation. In our work, we propose a novel Intent-Aware Cross Attention for next POI Recommendation (ICARec) to provide intent insight into the user’s short-term trajectory for better and accurate recommendation. We not only modeling the long/short-term user preference by segmented check-in sequences, but also design an intent cross-attention module which enables interactions between the user’s intent and their most recent check-ins. In addition, to enhance the pairwise independence of intents, we adopt an auxiliary loss function to disentangle intents, endowing them with greater interpretability by ensuring that each intent is represented as an independent and distinct construct within the model. Extensive experiments on two real-world datasets validate the superior performance of our model.

Graph-based social recommendation improves the prediction accuracy of recommendation by leveraging high-order neighboring information contained in social relations. However, most of them ignore the problem that social relations can be noisy for recommendation. Several studies attempt to tackle this problem by performing social graph denoising, but they suffer from 1) adaptability issues for other graph-based social recommendation models and 2) insufficiency issues for user social representation learning. To address the limitations, we propose a model-agnostic graph denoising module (denoted as MADM) which works as a plug-and-play module to provide refined social structure for base models. Meanwhile, to propel user social representations to be minimal and sufficient for recommendation, MADM further employs mutual information maximization (MIM) between user social representations and the interaction graph and realizes two ways of MIM: contrastive learning and forward predictive learning. We provide theoretical insights and guarantees from the perspectives of Information Theory and Multi-view Learning to explain its rationality. Extensive experiments on three real-world datasets demonstrate the effectiveness of MADM.

Deep learning-based recommender systems have achieved remarkable success in recent years. However, these methods usually heavily rely on labeled data (i.e., user-item interactions), suffering from problems such as data sparsity and cold-start. Self-supervised learning, an emerging paradigm that extracts information from unlabeled data, provides insights into addressing these problems. Specifically, contrastive self-supervised learning, due to its flexibility and promising performance, has attracted considerable interest and recently become a dominant branch in self-supervised learning-based recommendation methods. In this survey, we provide an up-to-date and comprehensive review of current contrastive self-supervised learning-based recommendation methods. Firstly, we propose a unified framework for these methods. We then introduce a taxonomy based on the key components of the framework, including view generation strategy, contrastive task, and contrastive objective. For each component, we provide detailed descriptions and discussions to guide the choice of the appropriate method. Finally, we outline open issues and promising directions for future research.

User cold-start recommendation is one of the most challenging problems that limit the effectiveness of recommender systems. Meta-learning-based methods are introduced to address this problem by learning initialization parameters for cold-start tasks. Recent studies attempt to enhance the initialization methods. They first represent each task by the cold-start user and interacted items. Then they distinguish tasks based on the task relevance to learn adaptive initialization. However, this manner is based on the assumption that user preferences can be reflected by the interacted items saliently, which is not always true in reality. In addition, we argue that previous approaches suffer from their adaptive framework (e.g., adaptive initialization), which reduces the adaptability in the process of transferring meta-knowledge to personalized RSs. In response to the issues, we propose a task-difficulty-aware meta-learning with adaptive update strategies (TDAS) for user cold-start recommendation. First, we design a task difficulty encoder, which can represent user preference salience, task relevance, and other task characteristics by modeling task difficulty information. Second, we adopt a novel framework with task-adaptive local update strategies by optimizing the initialization parameters with task-adaptive per-step and per-layer hyperparameters. Extensive experiments based on three real-world datasets demonstrate that our TDAS outperforms the state-of-the-art methods. The source code is available at https://github.com/XuHao-bit/TDAS.

Typhoons, formidable natural phenomena, typically unleash a trail of destruction, inflicting severe wind damage, floods, and even triggering tsunamis in coastal regions. Therefore, accurate prediction of typhoon intensity carries immense significance, both in theory and practical applications. Convolutional Neural Networks (CNNs) have shown superior capability in modelling spatial and temporal data. However, conventional convolutional operators are very sensitive to rotations, and is not good at modelling hierarchical structures. Since typhoon intensity data are typical spatiotemporal data that possess a distinctive rotational structure and a multi-layered hierarchy, conventional CNN based models fall short in capturing high-order spatial information, thereby limiting the accuracy of predictions. To address the limitations of conventional convolutional operators, we introduce a typhoon intensity prediction model called Multi-scale Spin Convolutional Neural Network (MS-CNN). Our model rotates the convolution kernel approximately to better model the rotating horizontal structure of the typhoons. Additionally, it employs a fine-grained multi-scale architecture to capture the hierarchical structure of typhoons. We conducted a series of comprehensive experiments using typhoon track dataset in the Western North Pacific. The results demonstrate that the superior performance of our MS-CNN model compared to other state-of-the-art models, marking a substantial advancement in typhoon intensity prediction tasks.

Textual reviews have been widely utilized in recommender systems because text reviews contain rich user preference information. Recent studies have increasingly incorporated textual reviews into user-item graphs as auxiliary information to learn node representations, notably enhancing recommendation performance. However, most existing review-based recommendations could not effectively exploit correlations between ratings and reviews, and suffer from noisy interactions which are further amplified during neighborhood aggregation. To address such limitations, we propose a new graph contrastive learning model for review-based recommendations in this paper. We construct a user-item graph view using both ratings and reviews. In addition, we design two graph views with ratings and reviews, respectively. Through contrastive learning based on these three views, our model manages to generate rich supervision signals for both user and item nodes. Our approach effectively explores intrinsic correlations between heterogeneous rating and review data, which enhance the robustness against interaction noises. A comprehensive experimental study on five benchmark datasets demonstrates that our model outperforms state-of-the-art methods.

Most existing collaborative filtering-based recommendation models only use a single type of interactive data. However, in real-world scenarios, there is more than one type of user interaction behavior. Although numerous recommendation models already use multi-behavior to make recommendation, there are still two limitations for them: i) the scarcity of supervised interaction data under the target behavior; ii) unable to obtain high-order collaborative relation due to over-smoothing. In order to solve the above problems, we propose a multi-behavior recommendation model, D ual C ontrastive L earning (DCL), based on contrastive learning. In DCL, a graph neural network is used to learn the local representations of users and items under multiple behaviors, while a hypergraph neural network is used to capture the global item dependency. In particular, we design two contrastive learning strategies to reinforce the quality of representations. Behavior-wise contrastive learning captures the knowledge that can be transferred between different behaviors. As well as item-wise contrastive learning captures the dependencies between item representations from different perspectives. The effectiveness of our method is verified on three real-world datasets, and the recommendation performance of DCL is superior to the various state-of-the-art recommendation models. At the same time, we also verify the robustness of our model for users with sparse interactions.

Predicting user behavior based on their interactions poses a great challenge in the realm of sequential recommendation. This challenge primarily stems from the erratic and periodic nature of user behavior. Recent approaches leverage attention mechanisms to capture users’ long- and short-term interests by analyzing their past interactions, and have achieved notable improvements of recommendation performance. However, none of these methods take into account semantic subsequences, which are indicative of a user’s focused activity over a specific period within the sequence. Incorporating semantic subsequences can significantly enhance the attention mechanism’s ability to capture users’ short-term interests.In this paper, we introduce the Semantic Subsequences Recognizer for Sequential Recommendation (SSR4Rec), comprising two distinct modules tailored for modeling long-term and short-term interests, respectively. For short-term interest modeling, we present a novel LSTM network that intelligently identifies semantic subsequences within sequences. It then extracts semantically coherent subsequences to create a representation of short-term interests. Meanwhile, for long-term interest modeling, we employ a modified BERT model to generate long-term interest representations. Our extensive experiments on three benchmark datasets demonstrates that SSR4Rec outperforms other state-of-the-art sequential models such as Locker and STOSA.

Graph Convolutional Networks (GCNs) have been widely used for collaborative filtering, due to their effectiveness in exploiting high-order collaborative signals. However, two issues have not been well addressed by existing studies. First, usually only one kind of information is utilized, i.e., user preference in user-item graphs or item dependency in item-item graphs. Second, they usually adopt static graphs, which cannot retain the temporal evolution of the information. These can limit the recommendation quality. To address these limitations, we propose to mine three kinds of information (user preference, item dependency, and user behavior similarity) and their temporal evolution by constructing multiple discrete dynamic heterogeneous graphs (i.e., a user-item dynamic graph, an item-item dynamic graph, and a user-subseq dynamic graph) from interaction data. A novel network (PDGCN) is proposed to learn the representations of users and items in these dynamic graphs. Moreover, we designed a structural neighbor aggregation module with novel pooling and convolution operations to aggregate the features of structural neighbors. We also design a temporal neighbor aggregation module based on self-attention mechanism to aggregate the features of temporal neighbors. We conduct extensive experiments on four real-world datasets. The results indicate that our approach outperforms several competing methods in terms of Hit Ratio (HR) and Normalized Discounted Cumulative Gain (NDCG). Dynamic graphs are also shown to be effective in improving recommendation performance.