Search

The article focuses on the challenge of conducting effective Privacy-Preserving Cross-Domain Recommendation (PPCDR) due to rising privacy concerns and the limited success of existing models in capturing heterogeneity across multiple domains. It mentions the proposed PPCDR approach introduces a federated graph learning method that combines global and local user preferences across domains while ensuring privacy through differential privacy techniques.

The article focuses on understanding diversity in session-based recommender systems (SBRSs) and examining the relationship between recommendation accuracy and diversity in these systems. Topics include the performance of representative SBRSs concerning both accuracy and diversity, the complex relationship between accuracy and diversity, and the factors influencing diversity in SBRSs.

The article focuses on the development of a Multi-view Enhanced Graph Attention Network (MEGAN) for session-based music recommendation, addressing the limitations of traditional music recommender systems that primarily rely on user interactions. Topics include the challenges posed by the increasing availability of digital music content, the limitations of traditional music recommender systems, and the changing patterns of users' music listening behavior.

The article focuses on the challenges and emerging expectations regarding the trustworthiness of recommendation and search systems, particularly in adapting to various use cases and ensuring robustness, interpretability, security, privacy, and fairness. It introduces a special section dedicated to novel research in this field, with a focus on promoting responsible AI applications and advancing techniques for the wider public.

The article centers on developing an automatic quotation recommendation system for online conversations, emphasizing a fusion of semantic and topic-based modeling techniques. It explores the challenges in recommending relevant quotations, validates the topic-based recommendation approach, and conducts extensive experiments to analyze recommendation difficulty and stability.

The article focuses on addressing biases in recommender systems caused by system-induced and user-induced factors. It introduces the concept of using a randomized dataset to mitigate system-induced biases and proposes a new theoretical framework to optimize the upper bound of an ideal objective function for debiasing. It also presents a novel method called "debiasing approximate upper bound (DUB)" and validates its effectiveness through extensive experiments on public and real product datasets.

Recommender systems relieve users from cognitive overloading by predicting preferred items for users. Due to the complexity of interactions between users and items, graph neural networks (GNN) use graph structures to effectively model user–item interactions. However, existing GNN approaches have the following limitations: (1) User reviews are not adequately modeled in graphs. Therefore, user preferences and item properties that are described in user reviews are lost for modeling users and items; and (2) GNNs assume deterministic relations between users and items, which lack the stochastic modeling to estimate the uncertainties in neighbor relations. To mitigate the limitations, we build tripartite graphs to model user reviews as nodes that connect with users and items. We estimate neighbor relations with stochastic variables and propose a Bayesian graph attention network (i.e., ContGraph) to accurately predict user ratings. ContGraph incorporates the prior knowledge of user preferences to regularize the posterior inference of attention weights. Our experimental results show that ContGraph significantly outperforms 13 state-of-the-artmodels and improves the best performing baseline (i.e., ANR) by 5.23% on 25 datasets in the five-core version. Moreover, we show that correctly modeling the semantics of user reviews in graphs can help express the semantics of users and items. [ABSTRACT FROM AUTHOR]

Recommender models are hard to evaluate, particularly under offline setting. In this article, we provide a comprehensive and critical analysis of the data leakage issue in recommender system offline evaluation. Data leakage is caused by not observing global timeline in evaluating recommenders e.g., train/test data split does not follow global timeline. As a result, a model learns from the user-item interactions that are not expected to be available at the prediction time. We first show the temporal dynamics of user-item interactions along global timeline, then explain why data leakage exists for collaborative filtering models. Through carefully designed experiments, we show that all models indeed recommend future items that are not available at the time point of a test instance, as the result of data leakage. The experiments are conducted with four widely used baseline models—BPR, NeuMF, SASRec, and LightGCN, on four popular offline datasets—MovieLens- 25M, Yelp, Amazon-music, and Amazon-electronic, adopting leave-last-one-out data split.1 We further show that data leakage does impact models’ recommendation accuracy. Their relative performance orders thus become unpredictable with different amount of leaked future data in training. To evaluate recommendation systems in a realistic manner in offline setting, we propose a timeline scheme, which calls for a revisit of the recommendation model design. [ABSTRACT FROM AUTHOR]

The article presents the discussion on recommendation and search systems becoming indispensable means for helping web users. Topics include applications of such systems being multi-faceted containing targeted advertising, intelligent medical assistant, and e-commerce; and robustness evaluating a model's performance consistency under various operating conditions like noisy data.

Graph neural networks (GNNs) have been widely applied in the recommendation tasks and have achieved very appealing performance. However, most GNN-based recommendation methods suffer from the problem of data sparsity in practice. Meanwhile, pre-training techniques have achieved great success in mitigating data sparsity in various domains such as natural language processing (NLP) and computer vision (CV). Thus, graph pre-training has the great potential to alleviate data sparsity in GNN-based recommendations. However, pre-training GNNs for recommendations faces unique challenges. For example, user-item interaction graphs in different recommendation tasks have distinct sets of users and items, and they often present different properties. Therefore, the successful mechanisms commonly used in NLP and CV to transfer knowledge from pre-training tasks to downstream tasks such as sharing learned embeddings or feature extractors are not directly applicable to existing GNN-based recommendations models. To tackle these challenges, we delicately design an adaptive graph pre-training framework for localized collaborative filtering (ADAPT). It does not require transferring user/item embeddings, and is able to capture both the common knowledge across different graphs and the uniqueness for each graph simultaneously. Extensive experimental results have demonstrated the effectiveness and superiority of ADAPT. [ABSTRACT FROM AUTHOR]

As recommender systems become increasingly important in daily human decision-making, users are demanding convincing explanations to understand why they get the specific recommendation results. Although a number of explainable recommender systems have recently been proposed, there still lacks an understanding of what users really need in a recommendation explanation. The actual reason behind users' intention to examine and consume (e.g., click and watch a movie) can be the window to answer this question and is named as self-explanation in this work. In addition, humans usually make recommendations accompanied by explanations, but there remain fewer studies on how humans explain and what we can learn from humangenerated explanations. To investigate these questions, we conduct a novel multi-role, multi-session user study inwhich users interact with multiple types of system-generated explanations as well as human-generated explanations, namely peer-explanation. During the study, users' intentions, expectations, and experiences are tracked in several phases, including before and after the users are presented with an explanation and after the content is examined. Through comprehensive investigations, three main findings have been made: First, we observe not only the positive but also the negative effects of explanations, and the impact varies across different types of explanations. Moreover, human-generated explanation, peer-explanation, performs better in increasing user intentions and helping users to better construct preferences, which results in better user satisfaction. Second, based on users' self-explanation, the information accuracy is measured and found to be a major factor associated with user satisfaction. Some other factors, such as unfamiliarity and similarity, are also discovered and summarized. Third, through annotations of the information aspects used in the human-generated selfexplanation and peer-explanation, patterns of how humans explain are investigated, including what information and how much information is utilized. In addition, based on the findings, a human-inspired explanation approach is proposed and found to increase user satisfaction, revealing the potential improvement of further incorporating more human patterns in recommendation explanations. These findings have shed light on the deeper understanding of the recommendation explanation and further research on its evaluation and generation. Furthermore, the collected data, including human-generated explanations by both the external peers and the users' selves, will be released to support future research works on explanation evaluation. [ABSTRACT FROM AUTHOR]

Nowadays, most online services are hosted onmulti-stakeholdermarketplaces, where consumers and producers may have different objectives. Conventional recommendation systems, however, mainly focus on maximizing consumers' satisfaction by recommending the most relevant items to each individual. This may result in unfair exposure of items, thus jeopardizing producer benefits. Additionally, they do not care whether consumers from diverse demographic groups are equally satisfied. To address these limitations, we propose a multi-objective optimization framework for fairness-aware recommendation, Multi-FR, that adaptively balances accuracy and fairness for various stakeholders with Pareto optimality guarantee. We first propose four fairness constraints on consumers and producers. In order to train the whole framework in an end-to-end way, we utilize the smooth rank and stochastic ranking policy to make these fairness criteria differentiable and friendly to back-propagation. Then, we adopt themultiple gradient descent algorithm to generate a Pareto set of solutions, from which the most appropriate one is selected by the Least Misery Strategy. The experimental results demonstrate that Multi-FR largely improves recommendation fairness on multiple stakeholders over the state-of-the-art approaches while maintaining almost the same recommendation accuracy. The training efficiency study confirms our model's ability to simultaneously optimize different fairness constraints for many stakeholders efficiently. [ABSTRACT FROM AUTHOR]

News recommender systems have become an effective manner to help users make decisions by suggesting the potential news that users may click and read, which has shown the proliferation nowadays. Many representative algorithms made great efforts to discover users' preferences from the histories for triggering news recommendations. However, there exist some limitations due to the following two main issues. First, they mainly rely on the sufficient user data, which cannot well capture users' temporal interests with very limited records. Second, always perceiving users' histories for recommendation may ignore some important news (e.g., breaking news). In this article, we propose a novel Multi-factors Fusion model for news recommendation by integrating both user-dependent preference effect and user-independent timeliness effect together. First, to track the preference of a certain user, we decompose her reading history into two user-related factors, including the long-term habit and the short-term interest. Specifically, we extract her persistent habit by exploring the category effect of news that she focuses on from her whole records. Then, we characterize her temporary interests by proposing a recurrent neural network of analyzing the homogeneous relations between her latest clicked news and the candidate ones. Second, to describe the user-independent news timeliness effect, we propose a novel survival analysis model to estimate the instantaneous click probability of a certain news as the occurring probability of an event, where much sensational news tends to be picked out. Last, we fuse all effects to determine the probability of a user clicking on a certain news under the independent event assumption. We conduct extensive experiments on two real-world datasets. Experimental results demonstrate that our model can generate better news recommendations on both general scenario and cold-start scenario. [ABSTRACT FROM AUTHOR]

Traditional recommendation systems are faced with two long-standing obstacles, namely data sparsity and cold-start problems, which promote the emergence and development of Cross-Domain Recommendation (CDR). The core idea of CDR is to leverage information collected from other domains to alleviate the two problems in one domain. Since the early 2010s, many efforts have been engaged for cross-domain recommendation. Recently, with the development of deep learning and neural networks, a large number of methods have emerged. However, there is a limited number of systematic surveys on CDR, especially regarding the latest proposed methods as well as the recommendation scenarios and recommendation tasks they address. In this survey article, we first proposed a two-level taxonomy of cross-domain recommendation that classifies different recommendation scenarios and recommendation tasks. We then introduce and summarize existing cross-domain recommendation approaches under different recommendation scenarios in a structured manner. We also organize datasets commonly used. We conclude this survey by providing several potential research directions about this field. [ABSTRACT FROM AUTHOR]

In recommender systems, top-N recommendation is an important task with implicit feedback data. Although the recent success of deep learning largely pushes forward the research on top-N recommendation, there are increasing concerns on appropriate evaluation of recommendation algorithms. It therefore is important to study how recommendation algorithms can be reliably evaluated and thoroughly verified. This work presents a large-scale, systematic study on six important factors from three aspects for evaluating recommender systems. We carefully select 12 top-N recommendation algorithms and eight recommendation datasets. Our experiments are carefully designed and extensively conducted with these algorithms and datasets. In particular, all the experiments in our work are implemented based on an open sourced recommendation library, Recbole [139], which ensures the reproducibility and reliability of our results. Based on the large-scale experiments and detailed analysis, we derive several key findings on the experimental settings for evaluating recommender systems. Our findings show that some settings can lead to substantial or significant differences in performance ranking of the compared algorithms. In response to recent evaluation concerns, we also provide several suggested settings that are specially important for performance comparison. [ABSTRACT FROM AUTHOR]

This article presents a novelmodel named Adversarial Auto-encoder Domain Adaptation to handle the recommendation problem under cold-start settings. Specifically, we divide the hypergraph into two hypergraphs, i.e., a positive hypergraph and a negative one. Below, we adopt the cold-start user recommendation for illustration. After achieving positive and negative hypergraphs, we apply hypergraph auto-encoders to them to obtain positive and negative embeddings of warm users and items. Additionally, we employ a multi-layer perceptron to get warm and cold-start user embeddings called regular embeddings. Subsequently, for warm users, we assign positive and negative pseudo-labels to their positive and negative embeddings, respectively, and treat their positive and regular embeddings as the source and target domain data, respectively. Then, we develop a matching discriminator to jointly minimize the classification loss of the positive and negative warm user embeddings and the distribution gap between the positive and regular warm user embeddings. In this way, warm users' positive and regular embeddings are connected. Since the positive hypergraph maintains the relations between positive warm user and item embeddings, and the regular warm and cold-start user embeddings follow a similar distribution, the regular cold-start user embedding and positive item embedding are bridged to discover their relationship. The proposed model can be easily extended to handle the coldstart item recommendation by changing inputs. We perform extensive experiments on real-world datasets for both cold-start user and cold-start item recommendations. Promising results in terms of precision, recall, normalized discounted cumulative gain, and hit rate verify the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

Personalized news recommendation is important for users to find interesting news information and alleviate information overload. Although it has been extensively studied over decades and has achieved notable success in improving user experience, there are still many problems and challenges that need to be further studied. To help researchers master the advances in personalized news recommendation, in this article, we present a comprehensive overview of personalized news recommendation. Instead of following the conventional taxonomy of news recommendation methods, in this article, we propose a novel perspective to understand personalized news recommendation based on its core problems and the associated techniques and challenges. We first review the techniques for tackling each core problem in a personalized news recommender system and the challenges they face. Next, we introduce the public datasets and evaluation methods for personalized news recommendation. We then discuss the key points on improving the responsibility of personalized news recommender systems. Finally, we raise several research directions that are worth investigating in the future. This article can provide up-to-date and comprehensive views on personalized news recommendation. We hope this article can facilitate research on personalized news recommendation as well as related fields in natural language processing and data mining. [ABSTRACT FROM AUTHOR]

To characterize complex and heterogeneous side information in recommender systems, the heterogeneous information network (HIN) has shown superior performance and attracted much research attention. In HIN, the rich entities, relations, and paths can be utilized to model the correlations of users and items; such a task setting is often called HIN-based recommendation. Although HIN provides a general approach to modeling rich side information, it lacks special consideration on the goal of the recommendation task. The aggregated context from the heterogeneous graph is likely to incorporate irrelevant information, and the learned representations are not specifically optimized according to the recommendation task. Therefore, there is a need to rethink how to leverage the useful information from HIN to accomplish the recommendation task. To address the above issues, we propose a Curriculum pre-training based HEterogeneous Subgraph Transformer (called CHEST) with new data characterization, representation model, and learning algorithm. Specifically, we consider extracting useful information from HIN to compose the interaction-specific heterogeneous subgraph, containing highly relevant context information for recommendation. Then, we capture the rich semantics (e.g., graph structure and path semantics) within the subgraph via a heterogeneous subgraph Transformer, where we encode the subgraph into multi-slot sequence representations. Besides, we design a curriculum pre-training strategy to provide an elementary-to-advanced learning process. The elementary course focuses on capturing local context information within the subgraph, and the advanced course aims to learn global context information. In this way, we gradually capture useful semantic information from HIN for modeling user-item interactions. Extensive experiments conducted on four real-world datasets demonstrate the superiority of our proposed method over a number of competitive baselines, especially when only limited training data is available. [ABSTRACT FROM AUTHOR]

Medicine recommendation systems target to recommend a set of medicines given a set of symptoms which play a crucial role in assisting doctors in their daily clinics. Existing approaches are either rule-based or supervised. However, the former heavily relies on expert labeling, which is time-consuming and costly to collect, and the latter suffers from the data sparse problem. To automate medicine recommendation on sparse data, we propose MedRec, which introduces two graphs in modeling: (1) a knowledge graph connecting diseases, medicines, symptoms, and examinations; (2) an attribute graph connecting medicines via shared attributes and molecular structures. These two graphs enhance the connectivity between symptoms and medicines, which thus alleviate the data sparse problem. By learning the interrelationship between diseases, medicines, symptoms and examinations and the inner relationship within medicine, we can acquire unified embedding representations of symptoms and medicines which can be used in medicine recommendation. The experimental results show that the proposed model outperforms state-of-the-art methods. In addition, we find that these two tasks: learning graph representation and medical recommendation can benefit each other. [ABSTRACT FROM AUTHOR]

The sequential recommendation (also known as the next-item recommendation), which aims to predict the following item to recommend in a session according to users’ historical behavior, plays a critical role in improving session-based recommender systems. Most of the existing deep learning-based approaches utilize the recurrent neural network architecture or self-attention to model the sequential patterns and temporal influence among a user’s historical behavior and learn the user’s preference at a specific time. However, these methods have two main drawbacks. First, they focus on modeling users’ dynamic states from a user-centric perspective and always neglect the dynamics of items over time. Second, most of them deal with only the first-order user-item interactions and do not consider the high-order connectivity between users and items, which has recently been proved helpful for the sequential recommendation. To address the above problems, in this article, we attempt to model user-item interactions by a bipartite graph structure and propose a new recommendation approach based on a Position-enhanced and Time-aware Graph Convolutional Network (PTGCN) for the sequential recommendation. PTGCN models the sequential patterns and temporal dynamics between user-item interactions by defining a position-enhanced and time-aware graph convolution operation and learning the dynamic representations of users and items simultaneously on the bipartite graph with a self-attention aggregator. Also, it realizes the high-order connectivity between users and items by stacking multi-layer graph convolutions. To demonstrate the effectiveness of PTGCN, we carried out a comprehensive evaluation of PTGCN on three real-world datasets of different sizes compared with a few competitive baselines. Experimental results indicate that PTGCN outperforms several state-of-the-art sequential recommendation models in terms of two commonly-used evaluation metrics for ranking. In particular, it can make a better trade-off between recommendation performance and model training efficiency, which holds great potential for online session-based recommendation scenarios in the future. [ABSTRACT FROM AUTHOR]

Sequential recommendation has become a trending research topic for its capability to capture dynamic user intents based on historical interaction sequence. To train a sequential recommendation model, it is a common practice to optimize the next-item recommendation task with a pairwise ranking loss. In this paper, we revisit this typical training method from the perspective of contrastive learning and find it can be taken as a specialized contrastive learning task conceptually and mathematically, named context-target contrast. Further, to leverage other self-supervised signals in user interaction sequences, we propose another contrastive learning task to encourage sequences after augmentation, as well as sequences with the same target item, to have similar representations, called context-context contrast. A general framework, ContraRec, is designed to unify the two kinds of contrast signals, leading to a holistic joint-learning framework for sequential recommendation with different contrastive learning tasks. Besides, various sequential recommendation methods (e.g., GRU4Rec, Caser, and BERT4Rec) can be easily integrated as the base sequence encoder in our ContraRec framework. Extensive experiments on three public datasets demonstrate that ContraRec achieves superior performance compared to state-of-the-art sequential recommendation methods. [ABSTRACT FROM AUTHOR]

Recommendation systems play an important role in alleviating the information overload issue. Generally, a recommendation model is trained to discern between positive (liked) and negative (disliked) instances for each user. However, under the open-world assumption, there are only positive instances but no negative instances from users’ implicit feedback, which poses the imbalanced learning challenge of lacking negative samples. To address this, two types of learning strategies have been proposed before, the negative sampling strategy and non-sampling strategy. The first strategy samples negative instances from missing data (i.e., unlabeled data), while the non-sampling strategy regards all the missing data as negative. Although learning strategies are known to be essential for algorithm performance, the in-depth comparison of negative sampling and non-sampling has not been sufficiently explored by far. To bridge this gap, we systematically analyze the role of negative sampling and non-sampling for implicit recommendation in this work. Specifically, we first theoretically revisit the objection of negative sampling and non-sampling. Then, with a careful setup of various representative recommendation methods, we explore the performance of negative sampling and nonsampling in different scenarios. Our results empirically show that although negative sampling has been widely applied to recent recommendation models, it is non-trivial for uniform sampling methods to show comparable performance to non-sampling learning methods. Finally, we discuss the scalability and complexity of negative sampling and non-sampling and present some open problems and future research topics that are worth being further explored. [ABSTRACT FROM AUTHOR]

Cross domain recommendation (CDR) is one popular research topic in recommender systems. This article focuses on a popular scenario for CDR where different domains share the same set of users but no overlapping items. The majority of recent methods have explored the shared-user representation to transfer knowledge across domains. However, the idea of shared-user representation resorts to learning the overlapped features of user preferences and suppresses the domain-specific features. Other works try to capture the domainspecific features by an MLP mapping but require heuristic human knowledge of choosing samples to train the mapping. In this article, we attempt to learn both features of user preferences in a more principled way. We assume that each user’s preferences in one domain can be expressed by the other one, and these preferences can be mutually converted to each other with the so-called equivalent transformation. Based on this assumption, we propose an equivalent transformation learner (ETL), which models the joint distribution of user behaviors across domains. The equivalent transformation in ETL relaxes the idea of shared-user representation and allows the learned preferences in different domains to preserve the domain-specific features as well as the overlapped features. Extensive experiments on three public benchmarks demonstrate the effectiveness of ETL compared with recent state-of-the-art methods. [ABSTRACT FROM AUTHOR]

We focus on Maximum Inner Product Search (MIPS), which is an essential problem in many machine learning communities. Given a query, MIPS finds the most similar items with the maximum inner products. Methods for Nearest Neighbor Search (NNS) which is usually defined on metric space do not exhibit the satisfactory performance for MIPS problem since inner product is a non-metric function. However, inner products exhibit many good properties compared with metric functions, such as avoiding vanishing and exploding gradients. As a result, inner product is widely used in many recommendation systems, which makes efficient Maximum Inner Product Search a key for speeding up many recommendation systems. Graph-based methods for NNS problem show the superiorities compared with other class methods. Each data point of the database is mapped to a node of the proximity graph. Nearest neighbor search in the database can be converted to route on the proximity graph to find the nearest neighbor for the query. This technique can be used to solve MIPS problem. Instead of searching the nearest neighbor for the query, we search the item with a maximum inner product with query on the proximity graph. In this article, we propose a reinforcement model to train an agent to search on the proximity graph automatically for MIPS problem if we lack the ground truths of training queries. If we know the ground truths of some training queries, our model can also utilize these ground truths by imitation learning to improve the agent’s searchability. By experiments, we can see that our proposed mode which combines reinforcement learning with imitation learning shows the superiorities over the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Incorporating knowledge graphs (KGs) into recommender systems to provide explainable recommendation has attracted much attention recently. The multi-hop paths in KGs can provide auxiliary facts for improving recommendation performance as well as explainability. However, existing studies may suffer from two major challenges: error propagation and weak explainability. Considering all paths between every user-item pair might involve irrelevant ones, which leads to error propagation of user preferences. Defining meta-paths might alleviate the error propagation, but the recommendation performance would heavily depend on the pre-defined meta-paths. Some recent methods based on graph convolution network (GCN) achieve better recommendation performance, but fail to provide explainability. To tackle the above problems, we propose a novel method named Knowledge-aware Reasoning with Graph Convolution Network (KR-GCN). Specifically, to alleviate the effect of error propagation, we design a transition-based method to determine the triple-level scores and utilize nucleus sampling to select triples within the paths between every user-item pair adaptively. To improve the recommendation performance and guarantee the diversity of explanations, user-item interactions and knowledge graphs are integrated into a heterogeneous graph, which is performed with the graph convolution network. A path-level self-attention mechanism is adopted to discriminate the contributions of different selected paths and predict the interaction probability, which improves the relevance of the final explanation. Extensive experiments conducted on three real-world datasets show that KR-GCN consistently outperforms several state-of-the-art baselines. And human evaluation proves the superiority of KR-GCN on explainability. [ABSTRACT FROM AUTHOR]

Item-based collaborative filtering (ICF) enjoys the advantages of high recommendation accuracy and ease in online penalization and thus is favored by the industrial recommender systems. ICF recommends items to a target user based on their similarities to the previously interacted items of the user. Great progresses have been achieved for ICF in recent years by applying advanced machine learning techniques (e.g., deep neural networks) to learn the item similarity from data. The early methods simply treat all the historical items equally and recently proposed methods attempt to distinguish the different importance of historical items when recommending a target item. Despite the progress, we argue that those ICF models neglect the diverse intents of users on adopting items (e.g., watching a movie because of the director, leading actors, or the visual effects). As a result, they fail to estimate the item similarity on a finer-grained level to predict the user’s preference to an item, resulting in sub-optimal recommendation. In this work, we propose a general feature-level attention method for ICF models. The key of our method is to distinguish the importance of different factors when computing the item similarity for a prediction. To demonstrate the effectiveness of our method, we design a light attention neural network to integrate both item-level and feature-level attention for neural ICF models. It is model-agnostic and easy-to-implement. We apply it to two baseline ICF models and evaluate its effectiveness on six public datasets. Extensive experiments show the feature-level attention enhanced models consistently outperform their counterparts, demonstrating the potential of differentiating user intents on the feature-level for ICF recommendation models. [ABSTRACT FROM AUTHOR]

Session-based recommendation aims to generate recommendations merely based on the ongoing session, which is a challenging task. Previous methods mainly focus on modeling the sequential signals or the transition relations between items in the current session using RNNs or GNNs to identify user’s intent for recommendation. Such models generally ignore the dynamic connections between the local and global item transition patterns, although the global information is taken into consideration by exploiting the global-level pair-wise item transitions. Moreover, existing methods that mainly adopt the crossentropy loss with softmax generally face a serious over-fitting problem, harming the recommendation accuracy. Thus, in this article, we propose a Graph Co-Attentive Recommendation Machine (GCARM) for session-based recommendation. In detail, we first design a Graph Co-Attention Network (GCAT) to consider the dynamic correlations between the local and global neighbors of each node during the information propagation. Then, the item-level dynamic connections between the output of the local and global graphs are modeled to generate the final item representations. After that, we produce the prediction scores and design a Max Cross-Entropy (MCE) loss to prevent over-fitting. Extensive experiments are conducted on three benchmark datasets, i.e., Diginetica, Gowalla, and Yoochoose. The experimental results show that GCARM can achieve the state-of-the-art performance in terms of Recall and MRR, especially on boosting the ranking of the target item. [ABSTRACT FROM AUTHOR]

As a major component of strategic talent management, learning and development (L&D) aims at improving the individual and organization performances through planning tailored training for employees to increase and improve their skills and knowledge. While many companies have developed the learning management systems (LMSs) for facilitating the online training of employees, a long-standing important issue is how to achieve personalized training recommendations with the consideration of their needs for future career development. To this end, in this article, we present a focused study on the explainable personalized online course recommender system for enhancing employee training and development. Specifically, we first propose a novel end-to-end hierarchical framework, namely Demand-aware Collaborative Bayesian Variational Network (DCBVN), to jointly model both the employees’ current competencies and their career development preferences in an explainable way. In DCBVN, we first extract the latent interpretable representations of the employees’ competencies from their skill profiles with autoencoding variational inference based topic modeling. Then, we develop an effective demand recognition mechanism for learning the personal demands of career development for employees. In particular, all the above processes are integrated into a unified Bayesian inference view for obtaining both accurate and explainable recommendations. Furthermore, for handling the employees with sparse or missing skill profiles, we develop an improved version of DCBVN, called the Demand-aware Collaborative Competency Attentive Network (DCCAN) framework, by considering the connectivity among employees. In DCCAN, we first build two employee competency graphs from learning and working aspects. Then, we design a graph-attentive network and a multi-head integration mechanism to infer one’s competency information from her neighborhood employees. Finally, we can generate explainable recommendation results based on the competency representations. Extensive experimental results on realworld data clearly demonstrate the effectiveness and the interpretability of both of our frameworks, as well as their robustness on sparse and cold-start scenarios. [ABSTRACT FROM AUTHOR]

The article presents the discussion on representing the relationships among data objects and graph-structure data being ubiquitous in real-world applications. Topics include organized as graphs, graph technologies attracted increasing attention from IR community; and pretraining personal word embeddings for each user in preserving word semantics and user interests.

The broad adoption of electronic health records (EHRs) has led to vast amounts of data being accumulated on a patient's history, diagnosis, prescriptions, and lab tests. Advances in recommender technologies have the potential to utilize this information to help doctors personalize the prescribed medications. However, existing medication recommendation systems have yet to make use of all these information sources in a seamless manner, and they do not provide a justification on why a particular medication is recommended. In this work, we design a two-stage personalized medication recommender system called PREMIER that incorporates information from the EHR. We utilize the various weights in the system to compute the contributions from the information sources for the recommended medications. Our system models the drug interaction from an external drug database and the drug co-occurrence from the EHR as graphs. Experiment results on MIMIC-III and a proprietary outpatient dataset show that PREMIER outperforms state-of-the-art medication recommendation systems while achieving the best tradeoff between accuracy and drug-drug interaction. Case studies demonstrate that the justifications provided by PREMIER are appropriate and aligned to clinical practices. [ABSTRACT FROM AUTHOR]

Explicit and implicit knowledge about users and items have been used to describe complex and heterogeneous side information for recommender systems (RSs). Many existing methods use knowledge graph embedding (KGE) to learn the representation of a user-item knowledge graph (KG) in low-dimensional space. In this article, we propose a lightweight end-to-end joint learning framework for fusing the tasks of KGE and RSs at the model level. Our method proposes a lightweight KG embedding method by using bidirectional bijection relation-type modeling to enable scalability for large graphs while using self-adaptive negative sampling to optimize negative sample generating. Our method further generates the integrated views for users and items based on relation-types to explicitly model users' preferences and items' features, respectively. Finally, we add virtual "recommendation" relations between the integrated views of users and items to model the preferences of users on items, seamlessly integrating RS with user-item KG over a unified graph. Experimental results on multiple datasets and benchmarks show that our method can achieve a better accuracy of recommendation compared with existing state-of-the-art methods. Complexity and runtime analysis suggests that our method can gain a lower time and space complexity than most of existing methods and improve scalability. [ABSTRACT FROM AUTHOR]

Social Recommender Systems (SRS) have attracted considerable attention since its accompanying service, social networks, helps increase user satisfaction and provides auxiliary information to improve recommendations. However, most existing SRS focus on social influence and ignore another essential social phenomenon, i.e., social homophily. Social homophily, which is the premise of social influence, indicates that people tend to build social relations with similar people and form influence propagation paths. In this article, we propose a generic framework Social PathExplorer (SPEX ) to enhance neural SRS. SPEX treats the neural recommendation model as a black box and improves the quality of recommendations by modeling the social recommendation task, the formation of social homophily, and their mutual effect in the manner of multi-task learning. We design a Graph Neural Network based component for influence propagation path prediction to help SPEX capture the rich information conveyed by the formation of social homophily. We further propose an uncertainty based task balancing method to set appropriate task weights for the recommendation task and the path prediction task during the joint optimization. Extensive experiments have validated that SPEX can be easily plugged into various state-of-the-art neural recommendation models and help improve their performance. [ABSTRACT FROM AUTHOR]

Fairness is a critical system-level objective in recommender systems that has been the subject of extensive recent research. A specific form of fairness is supplier exposure fairness, where the objective is to ensure equitable coverage of items across all suppliers in recommendations provided to users. This is especially important in multistakeholder recommendation scenarios where it may be important to optimize utilities not just for the end user but also for other stakeholders such as item sellers or producers who desire a fair representation of their items. This type of supplier fairness is sometimes accomplished by attempting to increase aggregate diversity to mitigate popularity bias and to improve the coverage of long-tail items in recommendations. In this article,we introduce FairMatch, a general graph-based algorithm thatworks as a post-processing approach after recommendation generation to improve exposure fairness for items and suppliers. The algorithm iteratively adds high-quality items that have low visibility or items from suppliers with low exposure to the users' final recommendation lists. A comprehensive set of experiments on two datasets and comparison with state-of-the-art baselines show that FairMatch, although it significantly improves exposure fairness and aggregate diversity, maintains an acceptable level of relevance of the recommendations. [ABSTRACT FROM AUTHOR]

A session-based recommender system (SBRS) captures users' evolving behaviors and recommends the next item by profiling users in terms of items in a session. User intent and user preference are two factors affecting his (her) decisions. Specifically, the former narrows the selection scope to some item types, while the latter helps to compare items of the same type. Most SBRSs assume one arbitrary user intent dominates a session when making a recommendation. However, this oversimplifies the reality that a session may involve multiple types of items conforming to different intents. In current SBRSs, items conforming to different user intents have cross-interference in profiling users for whom only one user intent is considered. Explicitly identifying and differentiating items conforming to various user intents can address this issue and model rich contextual information of a session. To this end, we design a framework modeling user intent and preference explicitly, which empowers the two factors to play their distinctive roles. Accordingly, we propose a key-array memory network (KA-MemNN) with a hierarchical intent tree to model coarse-to-fine user intents. The two-layer weighting unit (TLWU) in KA-MemNN detects user intents and generates intent-specific user profiles. Furthermore, the hierarchical semantic component (HSC) integrates multiple sets of intent-specific user profiles along with different user intent distributions to model a multi-intent user profile. The experimental results on real-world datasets demonstrate the superiority of KA-MemNN over selected state-of-the-art methods. [ABSTRACT FROM AUTHOR]

For present e-commerce platforms, it is important to accurately predict users' preference for a timely nextitem recommendation. To achieve this goal, session-based recommender systems are developed, which are based on a sequence of the most recent user-item interactions to avoid the influence raised from outdated historical records. Although a session can usually reflect a user's current preference, a local shift of the user's intention within the session may still exist. Specifically, the interactions that take place in the early positions within a session generally indicate the user's initial intention, while later interactions are more likely to represent the latest intention. Such positional information has been rarely considered in existing methods, which restricts their ability to capture the significance of interactions at different positions. To thoroughly exploit the positional information within a session, a theoretical framework is developed in this paper to provide an in-depth analysis of the positional information. We formally define the properties of forward-awareness and backward-awareness to evaluate the ability of positional encoding schemes in capturing the initial and the latest intention. According to our analysis, existing positional encoding schemes are generally forwardaware only, which can hardly represent the dynamics of the intention in a session. To enhance the positional encoding scheme for the session-based recommendation, a dual positional encoding (DPE) is proposed to account for both forward-awareness and backward-awareness. Based on DPE, we propose a novel Positional Recommender (PosRec) model with a well-designed Position-aware Gated Graph Neural Network module to fully exploit the positional information for session-based recommendation tasks. Extensive experiments are conducted on two e-commerce benchmark datasets, Yoochoose and Diginetica and the experimental results show the superiority of the PosRec by comparing it with the state-of-the-art session-based recommender models. [ABSTRACT FROM AUTHOR]

With the rapid development of online social recommendation system, substantial methods have been proposed. Unlike traditional recommendation system, social recommendation performs by integrating social relationship features, where there are two major challenges, i.e., early summarization and data sparsity. Thus far, they have not been solved effectively. In this article, we propose a novel social recommendation approach, namely Multi-Graph Heterogeneous Interaction Fusion (MG-HIF), to solve these two problems. Our basic idea is to fuse heterogeneous interaction features from multi-graphs, i.e., user--item bipartite graph and social relation network, to improve the vertex representation learning. A meta-path cross-fusion model is proposed to fuse multi-hop heterogeneous interaction features via discrete cross-correlations. Based on that, a social relation GAN is developed to explore latent friendships of each user. We further fuse representations from two graphs by a novel multi-graph information fusion strategy with attention mechanism. To the best of our knowledge, this is the first work to combine meta-path with social relation representation. To evaluate the performance of MG-HIF, we compare MG-HIF with seven states of the art over four benchmark datasets. The experimental results show that MG-HIF achieves better performance. [ABSTRACT FROM AUTHOR]

Personalized recommendation has become more and more important for users to quickly find relevant items. The key issue of the recommender system is how to model user preferences. Previous work mostly employed user historical data to learn users' preferences, but faced with the data sparsity problem. The prevalence of online social networks promotes increasing online discussion groups, and users in the same group often have similar interests and preferences. Therefore, it is necessary to integrate group information for personalized recommendation. The existing work on group-information-enhanced recommender systems mainly relies on the item information related to the group, which is not expressive enough to capture the complicated preference dependency relationships between group users and the target user. In this article, we solve the problem with the graph neural networks. Specifically, the relationship between users and items, the item preferences of groups, and the groups that users participate in are constructed as bipartite graphs, respectively, and the user preferences for items are learned end to end through the graph neural network. The experimental results on the Last.fm and Douban Movie datasets show that considering group preferences can improve the recommendation performance and demonstrate the superiority on sparse users compared [ABSTRACT FROM AUTHOR]

As the deep learning techniques have expanded to real-world recommendation tasks, many deep neural network based Collaborative Filtering (CF) models have been developed to project user-item interactions into latent feature space, based on various neural architectures, such as multi-layer perceptron, autoencoder, and graph neural networks. However, the majority of existing collaborative filtering systems are not well designed to handle missing data. Particularly, in order to inject the negative signals in the training phase, these solutions largely rely on negative sampling from unobserved user-item interactions and simply treating them as negative instances, which brings the recommendation performance degradation. To address the issues, we develop a Collaborative Reflection-Augmented Autoencoder Network (CRANet), that is capable of exploring transferable knowledge from observed and unobserved user-item interactions. The network architecture of CRANet is formed of an integrative structure with a reflective receptor network and an information fusion autoencoder module, which endows our recommendation framework with the ability of encoding implicit user’s pairwise preference on both interacted and non-interacted items. Additionally, a parametric regularizationbased tied-weight scheme is designed to perform robust joint training of the two-stage CRANet model. We finally experimentally validate CRANet on four diverse benchmark datasets corresponding to two recommendation tasks, to show that debiasing the negative signals of user-item interactions improves the performance as compared to various state-of-the-art recommendation techniques. [ABSTRACT FROM AUTHOR]

In recommender systems, it is essential to understand the underlying factors that affect user-item interaction. Recently, several studies have utilized disentangled representation learning to discover such hidden factors from user-item interaction data, which shows promising results. However, without any external guidance signal, the learned disentangled representations lack clear meanings, and are easy to suffer from the data sparsity issue. In light of these challenges, we study how to leverage knowledge graph (KG) to guide the disentangled representation learning in recommender systems. The purpose for incorporating KG is twofold, making the disentangled representations interpretable and resolving data sparsity issue. However, it is not straightforward to incorporate KG for improving disentangled representations, because KG has very different data characteristics compared with user-item interactions. We propose a novel Knowledge-guided Disentangled Representations approach (KDR) to utilizing KG to guide the disentangled representation learning in recommender systems. The basic idea, is to first learn more interpretable disentangled dimensions (explicit disentangled representations) based on structural KG, and then align implicit disentangled representations learned from user-item interaction with the explicit disentangled representations. We design a novel alignment strategy based on mutual information maximization. It enables the KG information to guide the implicit disentangled representation learning, and such learned disentangled representations will correspond to semantic information derived from KG. Finally, the fused disentangled representations are optimized to improve the recommendation performance. Extensive experiments on three real-world datasets demonstrate the effectiveness of the proposed model in terms of both performance and interpretability. [ABSTRACT FROM AUTHOR]

Recommendation systems are often evaluated based on user’s interactions that were collected from an existing, already deployed recommendation system. In this situation, users only provide feedback on the exposed items and they may not leave feedback on other items since they have not been exposed to them by the deployed system. As a result, the collected feedback dataset that is used to evaluate a new model is influenced by the deployed system, as a form of closed loop feedback. In this article, we show that the typical offline evaluation of recommender systems suffers from the so-called Simpson’s paradox. Simpson’s paradox is the name given to a phenomenon observed when a significant trend appears in several different sub-populations of observational data but disappears or is even reversed when these sub-populations are combined together. Our in-depth experiments based on stratified sampling reveal that a very small minority of items that are frequently exposed by the deployed system plays a confounding factor in the offline evaluation of recommendation systems. In addition, we propose a novel evaluation methodology that takes into account the confounder, i.e., the deployed system’s characteristics. Using the relative comparison of many recommendation models as in the typical offline evaluation of recommender systems, and based on the Kendall rank correlation coefficient, we show that our proposed evaluation methodology exhibits statistically significant improvements of 14% and 40% on the examined open loop datasets (Yahoo! and Coat), respectively, in reflecting the true ranking of systems with an open loop (randomised) evaluation in comparison to the standard evaluation. [ABSTRACT FROM AUTHOR]

Chatbots or conversational recommenders have gained increasing popularity as a new paradigm for Recommender Systems (RS). Prior work on RS showed that providing explanations can improve transparency and trust, which are critical for the adoption of RS. Their interactive and engaging nature makes conversational recommenders a natural platform to not only provide recommendations but also justify the recommendations through explanations. The recent surge of interest inexplainable AI enables diverse styles of justification, and also invites questions on howstyles of justification impact user perception. In this article, we explore the effect of "why" justifications and "why not" justifications on users' perceptions of explainability and trust. We developed and tested a movie-recommendation chatbot that provides users with different types of justifications for the recommended items. Our online experiment (n = 310) demonstrates that the "why" justifications (but not the "why not" justifications) have a significant impact on users' perception of the conversational recommender. Particularly, "why" justifications increase users' perception of system transparency, which impacts perceived control, trusting beliefs and in turn influences users' willingness to depend on the system's advice. Finally, we discuss the design implications for decision-assisting chatbots. [ABSTRACT FROM AUTHOR]

In the Internet of Things era, users are willing to personalize the joint behavior of their connected entities, i.e., smart devices and online service, by means of trigger-action rules such as "IF the entrance Nest security camera detects a movement, THEN blink the Philips Hue lamp in the kitchen." Unfortunately, the spread of new supported technologies makes the number of possible combinations between triggers and actions continuously growing, thus motivating the need of assisting users in discovering new rules and functionality, e.g., through recommendation techniques. To this end, we present HeyTAP2, a semantic Conversational Search and Recommendation (CSR) system able to suggest pertinent IF-THEN rules that can be easily deployed in different contexts starting from an abstract user's need. By exploiting a conversational agent, the user can communicate her current personalization intention by specifying a set of functionality at a high level, e.g., to decrease the temperature of a room when she left it. Stemming from this input, HeyTAP2 implements a semantic recommendation process that takes into account (a) the current user's intention, (b) the connected entities owned by the user, and (c) the user's long-term preferences revealed by her profile. If not satisfied with the suggestions, then the user can converse with the system to provide further feedback, i.e., a short-term preference, thus allowing HeyTAP2 to provide refined recommendations that better align with the original intention. We evaluate HeyTAP2 by running different offline experiments with simulated users and real-world data. First, we test the recommendation process in different configurations, and we show that recommendation accuracy and similarity with target items increase as the interaction between the algorithm and the user proceeds. Then, we compare HeyTAP2 with other similar baseline recommender systems. Results are promising and demonstrate the effectiveness of HeyTAP2 in recommending IF-THEN rules that satisfy the current personalization intention of the user. [ABSTRACT FROM AUTHOR]

Recent observational studies highlight the importance of considering the interactions between users in the group recommendation process, but to date their integration has been marginal. In this article, we propose a collaborative model based on the social interactions that take place in a web-based conversational group recommender system. The collaborative model allows the group recommender to implicitly infer the different roles within the group, namely, collaborative and leader user(s). Moreover, it serves as the basis of several novel collaboration-based consensus strategies that integrate both individual and social interactions in the group recommendation process. A live-user evaluation confirms that our approach accurately identifies the collaborative and leader users in a group and produces more effective recommendations. [ABSTRACT FROM AUTHOR]

In this article, we present MyrrorBot, a personal digital assistant implementing a natural language interface that allows the users to: (i) access online services, such as music, video, news, and food recommendations, in a personalized way, by exploiting a strategy for implicit user modeling called holistic user profiling; (ii) query their own user models, to inspect the features encoded in their profiles and to increase their awareness of the personalization process. Basically, the system allows the users to formulate natural language requests related to their information needs. Such needs are roughly classified in two groups: quantified self-related needs (e.g., Did I sleep enough? Am I extrovert?) and personalized access to online services (e.g., Play a song I like). The intent recognition strategy implemented in the platform automatically identifies the intent expressed by the user and forwards the request to specific services and modules that generate an appropriate answer that fulfills the query. In the experimental evaluation, we evaluated both qualitative (users' acceptance of the system, usability) as well as quantitative (time required to complete basic tasks, effectiveness of the personalization strategy) aspects of the system, and the results showed that MyrrorBot can improve the way people access online services and applications. This leads to a more effective interaction and paves theway for further development of our system. [ABSTRACT FROM AUTHOR]

Static recommendation methods like collaborative filtering suffer from the inherent limitation of performing real-time personalization for cold-start users. Online recommendation, e.g., multi-armed bandit approach, addresses this limitation by interactively exploring user preference online and pursuing the explorationexploitation (EE) trade-off. However, existing bandit-based methods model recommendation actions homogeneously. Specifically, they only consider the items as the arms, being incapable of handling the item attributes, which naturally provide interpretable information of user's current demands and can effectively filter out undesired items. In this work, we consider the conversational recommendation for cold-start users, where a system can both ask the attributes from and recommend items to a user interactively. This important scenario was studied in a recent work [54]. However, it employs a hand-crafted function to decide when to ask attributes or make recommendations. Such separate modeling of attributes and items makes the effectiveness of the system highly rely on the choice of the hand-crafted function, thus introducing fragility to the system. To address this limitation, we seamlessly unify attributes and items in the same arm space and achieve their EE trade-offs automatically using the framework of Thompson Sampling. Our Conversational Thompson Sampling (ConTS) model holistically solves all questions in conversational recommendation by choosing the arm with the maximal reward to play. Extensive experiments on three benchmark datasets show that ConTS outperforms the state-of-the-art methods Conversational UCB (ConUCB) [54] and Estimation--Action--Reflection model [27] in both metrics of success rate and average number of conversation turns. [ABSTRACT FROM AUTHOR]

We investigate the impact of popularity bias in false-positive metrics in the offline evaluation of recommender systems. Unlike their true-positive complements, false-positivemetrics reward systems thatminimize recommendations disliked by users. Our analysis is, to the best of our knowledge, the first to showthat false-positive metrics tend to penalise popular items, the opposite behavior of true-positive metrics--causing a disagreement trend between both types of metrics in the presence of popularity biases. We present a theoretical analysis of the metrics that identifies the reason that the metrics disagree and determines rare situations where the metrics might agree--the key to the situation lies in the relationship between popularity and relevance distributions, in terms of their agreement and steepness--two fundamental concepts we formalize. We then examine three well-known datasets using multiple popular true- and false-positive metrics on 16 recommendation algorithms. Specific datasets are chosen to allow us to estimate both biased and unbiased metric values. The results of the empirical study confirm and illustrate our analytical findings. With the conditions of the disagreement of the two types of metrics established, we then determine under which circumstances true-positive or false-positive metrics should be used by researchers of offline evaluation in recommender systems. [ABSTRACT FROM AUTHOR]

Sampling strategies have been widely applied in many recommendation systems to accelerate model learning from implicit feedback data. A typical strategy is to draw negative instances with uniform distribution, which, however, will severely affect a model's convergence, stability, and even recommendation accuracy. A promising solution for this problem is to over-sample the "difficult" (a.k.a. informative) instances that contribute more on training. But this will increase the risk of biasing the model and leading to non-optimal results. Moreover, existing samplers are either heuristic, which require domain knowledge and often fail to capture real "difficult" instances, or rely on a sampler model that suffers from low efficiency. To deal with these problems, we propose CoSam, an efficient and effective collaborative sampling method that consists of (1) a collaborative sampler model that explicitly leverages user-item interaction information in sampling probability and exhibits good properties of normalization, adaption, interaction information awareness, and sampling efficiency, and (2) an integrated sampler-recommender framework, leveraging the sampler model in prediction to offset the bias caused by uneven sampling. Correspondingly, we derive a fast reinforced training algorithm of our framework to boost the sampler performance and sampler-recommender collaboration. Extensive experiments on four real-world datasets demonstrate the superiority of the proposed collaborative sampler model and integrated sampler-recommender framework. [ABSTRACT FROM AUTHOR]

Sequential recommendation, such as next-basket recommender systems (NBRS), which model users' sequential behaviors and the relevant context/session, has recently attracted much attention from the research community. Existing session-based NBRS involve session representation and inter-basket relations but ignore their hybrid couplings with the intra-basket items, often producing irrelevant or similar items in the next basket. In addition, they do not predict next-baskets (more than one next basket recommended). Interactive recommendation further involves user feedback on the recommended basket. The existing work on next-item recommendation involves positive feedback on selected items but ignores negative feedback on unselected ones. Here, we introduce a new setting--interactive sequential basket recommendation, which iteratively predicts next baskets by learning the intra-/inter-basket couplings between items and both positive and negative user feedback on recommended baskets. A hierarchical attentive encoder-decoder model (HAEM) continuously recommends next baskets one after another during sequential interactions with users after analyzing the item relations both within a basket and between adjacent sequential baskets (i.e., intra-/inter-basket couplings) and incorporating the user selection and unselection (i.e., positive/negative) feedback on the recommended baskets to refine NBRS. HAEM comprises a basket encoder and a sequence decoder to model intra-/interbasket couplings and a prediction decoder to sequentially predict next-baskets by interactive feedback-based refinement. Empirical analysis shows that HAEM significantly outperforms the state-of-the-art baselines for NBRS and session-based recommenders for accurate and novel recommendation. We also show the effect of continuously refining sequential basket recommendation by including unselection feedback during interactive recommendation. [ABSTRACT FROM AUTHOR]

Product reviews can provide rich information about the opinions users have of products. However, it is nontrivial to effectively infer user preference and item characteristics from reviews due to the complicated semantic understanding. Existing methods usually learn features for users and items from reviews in single static fashions and cannot fully capture user preference and item features. In this article, we propose a neural review-based recommendation approach that aims to learn comprehensive representations of users/items under a three-tier attention framework. We design a review encoder to learn review features from words via a word-level attention, an aspect encoder to learn aspect features via a review-level attention, and a user/item encoder to learn the final representations of users/items via an aspect-level attention. In word- and review-level attentions, we adopt the context-aware mechanism to indicate importance of words and reviews dynamically instead of static attention weights. In addition, the attentions in the word and review levels are of multiple paradigms to learn multiple features effectively, which could indicate the diversity of user/item features. Furthermore, we propose a personalized aspect-level attention module in user/item encoder to learn the final comprehensive features. Extensive experiments are conducted and the results in rating prediction validate the effectiveness of our method. [ABSTRACT FROM AUTHOR]