GPS: Factorized group preference-based similarity models for sparse sequential recommendation

Predicting personalized sequential behavior is a key task for recommender systems. In order to predict user actions such as the next product to purchase, movie to watch, or place to visit, it is essential to take into account both long-term user preferences and sequential patterns (i.e., short-term dynamics). Matrix Factorization and Markov Chain methods have emerged as two separate but powerful paradigms for modeling the two respectively. Combining these ideas has led to unified methods that accommodate long-and short-term dynamics simultaneously by modeling pairwise user-item and item-item interactions. In spite of the success of such methods for tackling dense data, they are challenged by sparsity issues, which are prevalent in real-world datasets. In recent years, similarity-based methods have been proposed for (sequentially-unaware) item recommendation with promising results on sparse datasets. In this paper, we propose to fuse such methods with Markov Chains to make personalized sequential recommendations. We evaluate our method, Fossil, on a variety of large, real-world datasets. We show quantitatively that Fossil outperforms alternative algorithms, especially on sparse datasets, and qualitatively that it captures personalized dynamics and is able to make meaningful recommendations.

User preferences are usually dynamic in real-world recommender systems, and a user»s historical behavior records may not be equally important when predicting his/her future interests. Existing recommendation algorithms -- including both shallow and deep approaches -- usually embed a user»s historical records into a single latent vector/representation, which may have lost the per item- or feature-level correlations between a user»s historical records and future interests. In this paper, we aim to express, store, and manipulate users» historical records in a more explicit, dynamic, and effective manner. To do so, we introduce the memory mechanism to recommender systems. Specifically, we design a memory-augmented neural network (MANN) integrated with the insights of collaborative filtering for recommendation. By leveraging the external memory matrix in MANN, we store and update users» historical records explicitly, which enhances the expressiveness of the model. We further adapt our framework to both item- and feature-level versions, and design the corresponding memory reading/writing operations according to the nature of personalized recommendation scenarios. Compared with state-of-the-art methods that consider users» sequential behavior for recommendation, e.g., sequential recommenders with recurrent neural networks (RNN) or Markov chains, our method achieves significantly and consistently better performance on four real-world datasets. Moreover, experimental analyses show that our method is able to extract the intuitive patterns of how users» future actions are affected by previous behaviors.

Modeling user preferences from users' historical sequences is one of the core problems of sequential recommendation. Previous studies only considered transition patterns between user-items, ignoring transition patterns between item features and item-item interactions. Recently, there has been interest in integrating knowledge graphs as auxiliary information into sequential recommendation. Most of the existing methods deal with the heterogeneous information in the knowledge graph in a coarse-grained manner. We believe that fine-grained processing of information in knowledge graphs can help recommendation systems understand changes in user preferences, i.e. dividing hetero-geneous information into item-to-item relationships and item-to-attribute relationships. In this paper, we propose a dual-level self-attention network for sequential recommendation. Specifically, we divide the knowledge graph heterogeneous information about items into relation level and attribute level, representing item-item relationship and item-attribute relationship, respectively. Afterwards, the self-attention network is used to learn user preferences at dual-level, respectively. Then, the outputs of the above dual levels are integrated for next item recommendation. Based on extensive experiments on three real-world data sets, our model achieves significant improvements compared to state-of-the-art baseline methods.

Sequential recommendation aims at identifying the next item that is preferred by a user based on their behavioral history. Compared to conventional sequential models that leverage attention mechanisms and RNNs, recent efforts mainly follow two directions for improvement: multi-interest learning and graph convolutional aggregation. Specifically, multi-interest methods such as ComiRec and MIMN, focus on extracting different interests for a user by performing historical item clustering, while graph convolution methods including TGSRec and SURGE elect to refine user preferences based on multilevel correlations between historical items. Unfortunately, neither of them realizes that these two types of solutions can mutually complement each other, by aggregating multi-level user preference to achieve more precise multi-interest extraction for a better recommendation. To this end, in this paper, we propose a unified multi-grained neural model (named MGNM) via a combination of multi-interest learning and graph convolutional aggregation. Concretely, MGNM first learns the graph structure and information aggregation paths of the historical items for a user. It then performs graph convolution to derive item representations in an iterative fashion, in which the complex preferences at different levels can be well captured. Afterwards, a novel sequential capsule network is proposed to inject the sequential patterns into the multi-interest extraction process, leading to a more precise interest learning in a multi-grained manner. Experiments on three real-world datasets from different scenarios demonstrate the superiority of MGNM against several state-of-the-art baselines. The performance gain over the best baseline is up to 27.10% and 25.17% in terms of [email protected] and [email protected] respectively, which is one of the largest gains in recent development of sequential recommendation. Further analysis also demonstrates that MGNM is robust and effective at user preference understanding at multi-grained levels.

Most sequential recommendation (SR) systems employing graph neural networks (GNNs) only model a user's interaction sequence as a flat graph without hierarchy, overlooking diverse factors in the user's preference. Moreover, the timespan between interacted items is not sufficiently utilized by previous models, restricting SR performance gains. To address these problems, we propose a novel SR system employing a hierarchical graph neural network (HGNN) to model factorial user preferences. Specifically, a timespan-aware sequence graph (TSG) for the target user is first constructed with the timespan among interacted items. Next, all original nodes in TSG are softly clustered into factor nodes, each of which represents a certain factor of the user's preference. At last, all factor nodes' representations are used together to predict SR results. Our extensive experiments upon two datasets justify that our HGNN-based factorial user modeling obtains better SR performance than the state-of-the-art SR models.

Sequential recommender systems, as a specialized branch of recommender systems that can capture users’ dynamic preferences for more accurate and timely recommendations, have made significant progress. Recently, due to increasing concerns about user data privacy, some researchers have implemented federated learning for sequential recommendation, a.k.a., Federated Sequential Recommender Systems (FedSeqRecs), in which a public sequential recommender model is shared and frequently transmitted between a central server and clients to achieve collaborative learning. Although these solutions mitigate user privacy to some extent, they present two significant limitations that affect their practical usability: (1) They require a globally shared sequential recommendation model. However, in real-world scenarios, the recommendation model constitutes a critical intellectual property for platform and service providers. Therefore, service providers may be reluctant to disclose their meticulously developed models. (2) The communication costs are high as they correlate with the number of model parameters. This becomes particularly problematic as the current FedSeqRec will be inapplicable when sequential recommendation marches into a large language model era. To overcome the above challenges, this article proposes a parameter transmission-free federated sequential recommendation framework (PTF-FSR), which ensures both model and data privacy protection to meet the privacy needs of service providers and system users alike. Furthermore, since PTF-FSR only transmits prediction results under privacy protection, which are independent of model sizes, this new federated learning architecture can accommodate more complex and larger sequential recommendation models. Extensive experiments conducted on three widely used recommendation datasets, employing various sequential recommendation models from both ID-based and ID-free paradigms, demonstrate the effectiveness and generalization capability of our proposed framework. To facilitate future research in this direction, we release our code at https://github.com/hi-weiyuan/PTF-FSR .

Sequential recommendation aims to predict users' preferences in the future based on the interactions that they have done recently. Since users only interact with a small part of items, there is a problem of data sparsity in real-world datasets. Some studies leverage collaborative filtering (CF) to model collaborative relations to alleviate data sparsity. However, CF focuses on modeling first-order relations and ignores high-order collaborative relations. Meanwhile, most of the models assume the attributes of items are static, and only consider users' dynamics. In this paper, we propose Dual Sequential Recommendation with Collaborative Relations (DSRCR) which aggregates high-order collaborative features on the user-user graph and item-item graph via graph attention network as the input of dual sequential models and utilizes the attention-based recurrent neural network to model both user-side and item-side sequences. The contribution of different hidden states to user preferences or item attributes can be captured by the attention mechanism. Each user (item) is represented as the composition of the embedding with high-order collaborative features and the embedding that obtains from sequential models. Extensive experiments on three datasets demonstrate the superiority of our proposed model compared with other state-of-the-art baselines.

Transformer-based sequential recommendation (SR) models excel at modeling long-range dependencies, but suffer from high computational costs and catastrophic forgetting during continuous updates. Although continual learning has been applied to recommendation, existing methods gradually forget long-term user preferences and remain underexplored in SR. In this paper, we introduce Continual Sequential Transformer for Recommendation (CSTRec), which effectively adapt to current interests by leveraging preserved historical knowledge. Its core is Continual Sequential Attention (CSA), a linear attention tailored for continual SR, which partially retain historical knowledge without direct access to prior data. CSA features: (1) Cauchy-Schwarz Normalization to stabilize learning over time under uneven user interaction frequencies, and (2) Collaborative Interest Enrichment via shared, learnable interest pools to mitigate forgetting. We also introduce a new technique for new user adaptation by transferring historical knowledge from existing users with similar interests. Extensive experiments show CSTRec's superior performance in both knowledge retention and acquisition. Our code is available at https://github.com/Gyu-Seok0/CSTRec_WSDM26.

A point-of-interest(POI) recommendation aims to mine a user’s visiting history and find her/his potentially preferred places. The decision process when choosing a POI is complex and can be influenced by numerous factors, including personal preferences, geographical considerations, and user social relations. While latent factor models have been proven effective and are widely used for recommendations, adopting them to POI recommendations requires delicate consideration of the unique characteristics of location-based social networks (LBSNs). To this end, in this paper, we propose a joint convolution matrix factorization model, named the Review Geographical Social (ReGS) which strategically takes various factors into consideration. Specifically, this model captures geographical influences from a user’s check-in behaviour, and user social relations can be effectively leveraged in the recommendation model. The reviews information available on LBSNs could be related to a user’s check-in action, providing a unique opportunity for a POI recommendation. We model above three types of information under a unified POI recommendation framework based on convolution matrix factorization which integrates a convolutional neural network into a probability matrix factorization. Finally, we conduct a comprehensive performance evaluation for the ReGS using two real-world datasets collected from Foursquare. Experimental results show that the ReGS achieves significantly superior precision and recall rates to other state-of-the-art recommendation models.

Increasing multiple behavior recommendation models have achieved great successes. However, many models do not consider commonalities and differences between behaviors and data sparsity of the target behavior. This article proposes a novel multi-behavior recommendation model based on contrastive clustering learning (MBRCC). Specifically, the graph convolutional network (GCN) is employed to obtain the embeddings of users and items, respectively. Then, three kinds of tasks (including behavior-level embedding, instance-level embedding, and cluster-level embedding) are designed to optimize the embeddings of users and items. In behavior-level embedding, we design an adaptive parameter learning strategy to analyze the impact of auxiliary behaviors on the target behavior. Then, the embeddings of users for each behavior are weighted to obtain the final embeddings of users. In instance-level embedding, we employ contrastive learning to analyze the instances of user and item for mitigating the issue of data sparsity. In cluster-level embedding, we design a new cluster contrastive learning method to capture the similarity between groups of user and item. Finally, we combine these three tasks to improve the quality of the embeddings of users and items. We conduct extensive experiments on three real-world datasets and experimental results indicate that the MBRCC remarkably outperforms numerous existing recommendation models.

Time always exists in our lives and time data can easily be collected in a variety of applications. For example, when you purchase items online or click on an ad, the time at which you chose the item or clicked the ad is recorded. The analysis of time information can therefore be applied in various areas. It is important to note that user preferences change over time. For example, a person who watched animated TV shows in childhood will most likely switch to watching the news in adulthood. It is effective to incorporate such changes into recommender systems. In this paper, we propose an approach that predicts user preferences with consideration of preference changes by learning the order of purchase history in a recommender system. Our approach is composed of three steps. First, we obtain user features based on matrix factorization and purchasing time. Next, we use a Kalman filter to predict user preference vectors from user features. Finally, we generate a recommendation list, at which time we propose two types of recommendation methods using the predicted vectors. We then show through experiments using a real-world dataset that our approach outperforms competitive methods such as the first order Markov model.

Introduction: Recommendation systems are essential tools for facilitating user access to information and addressing the challenge of information overload. They operate by analyzing historical behaviors to learn user preferences. This study explores a hybrid recommendation approach that combines collaborative filtering methods: analyzing sequences of watched movies and leveraging movie ratings. The goal is to enhance recommendation accuracy by dynamically adjusting these methods based on specific use cases. Methods: The proposed method uses a weighted post-fusion technique to combine the two collaborative filtering approaches. For instance, potentially higher weights are assigned to the rating matrix when the majority of data includes ratings, while sequential data receives more emphasis when ratings are scarce. objective: to give vaid recommendations to users based on comprehensive datasets. Results: Technique could integrate sequential data with ratings effectively, capturing both short-term and long-term user preferences. The approach addresses limitations of standalone models, such as insufficient long-term preferences or a lack of rating considerations. method: various weights would be set based on use cases. For instance, in cases where we have the rating for the majority of classes, we will assign a higher weight to the rating matrix and in case where the rating is unavailable for the majority of cases, the higher weights might be assigned to the sequential dataset. Discussion: Integrating heterogeneous data sources within a unified recommendation framework leads to richer, more adaptive user modeling. The dynamic fusion approach not only personalizes results but also aligns recommendations with evolving user patterns. Future work may include learning adaptive weights automatically through meta-learning, incorporating metadata like genres or timestamps, and scaling to real-world datasets with millions of users and items. Conclusion: This hybrid methodology highlights the importance of combining multiple data sources to create more personalized and effective recommendations. The study provides a framework for adapting weights to align with varied use cases, paving the way for advancements in hybrid recommendation systems. conclusion: na

Matrix factorization (MF) has been proved to be an effective approach to build a successful recommender system. However, most current MF-based recommenders cannot obtain high prediction accuracy due to the sparseness of user-item matrix. Moreover, these methods suffer from the scalability issues when applying on large-scale real-world tasks. To tackle these issues, in this paper a social regularization method called TrustRSNMF is proposed that incorporates the social trust information of users in nonnegative matrix factorization framework. The proposed method integrates trust statements along with user-item ratings as an additional information source into the recommendation model to deal with the data sparsity and cold-start issues. In order to evaluate the effectiveness of the proposed method, a number of experiments are performed on two real-world datasets. The obtained results demonstrate significant improvements of the proposed method compared to state-of-the-art recommendation methods.

An increased interest in sequential recommendation has been observed in recent years. Many models have been proposed to leverage the sequential user-item interaction data, which includes those based on Markov Chain or recurrent neural networks. Most of these models are designed for the scenario where each historical record composed of single item. However, the records could be a subset of items (or session) such as music playlists and baskets in e-commerce applications. How to leverage the session structure to improve the effectiveness of the recommendation system is a challenge. To this end, we propose a MEmory-augmented Attention Network for Sequential recommendation (MEANS), to effectively recommend next items given the sequential session data. The most recent sessions are stored into external memory after a max-pooling operation. The long-term user preference are learned through an attention network which is stacked on the memory layer. Finally, the mixture of long-term and short-term preference is feeded into the prediction layer to make recommendations. Extensive experiments on four real datasets show that MEANS outperforms various state-of-the-art sequential recommendation models.

Collaborative filtering predicts user preferences by modeling the user's historical behavior. Matrix factorization plays an important role in collaborative filtering. Matrix factorization use dot product to predicts rating. Nevertheless, the dot product does not satisfy the triangle inequality, which may limit their expressiveness and lead to sub-optimal solutions. This paper combines metric learning and distance factorization. Firstly, convert the scoring matrix to a distance matrix. Then the distance matrix is factorization by metric learning to obtain the position of users and items in low-dimensional Euclidean space. User preferences and distance are negatively correlated, the user is closer to the favorite item and farther away from the unloved item, that is, the user's preference is reflected by the distance. Experiments conducted on real-world data sets have shown that compared with classical algorithm, our method significantly improves the accuracy of recommendations.