Efficient federated item similarity model for privacy-preserving recommendation

Previous federated recommender systems are based on traditional matrix factorization, which can improve personalized service but are vulnerable to gradient inference attacks. Most of them adopt model averaging to fit the data heterogeneity of federated recommender systems, requiring more training costs. To address privacy and efficiency, we propose an efficient federated item similarity model for the heterogeneous recommendation, called FedIS, which can train a global item-based collaborative filtering model to eliminate user feature dependencies. Specifically, we extend the neural item similarity model to the federated model, where each client only locally optimizes the shared item feature matrix. We then propose a fast-convergent federated aggregation method inspired by meta-learning to address heterogeneous user updates and accelerate the convergence of global training. Furthermore, we propose a two-stage perturbation method to protect both local training and transmission while reducing communication costs. Finally, extensive experiments on four real-world datasets validate that FedIS can provide more competitive performance on federated recommendations. Our proposed method also shows significant training efficiency with less performance degradation.     

Dynamic intent-aware iterative denoising network for session-based recommendation

Session-based recommendation aims to predict items that a user will interact with based on historical behaviors in anonymous sessions. It has long faced two challenges: (1) the dynamic change of user intents which makes user preferences towards items change over time; (2) the uncertainty of user behaviors which adds noise to hinder precise preference learning. They jointly preclude recommender system from capturing real intents of users. Existing methods have not properly solved these problems since they either ignore many useful factors like the temporal information when building item embeddings, or do not explicitly filter out noisy clicks in sessions. To tackle above issues, we propose a novel Dynamic Intent-aware Iterative Denoising Network (DIDN) for session-based recommendation. Specifically, to model the dynamic intents of users, we present a dynamic intent-aware module that incorporates item-aware, user-aware and temporal-aware information to learn dynamic item embeddings. A novel iterative denoising module is then devised to explicitly filter out noisy clicks within a session. In addition, we mine collaborative information to further enrich the session semantics. Extensive experimental results on three real-world datasets demonstrate the effectiveness of the proposed DIDN. Specifically, DIDN obtains improvements over the best baselines by 1.66%, 1.75%, and 7.76% in terms of P@20 and 1.70%, 2.20%, and 10.48% in terms of MRR@20 on all datasets.     

Group event recommendation based on graph multi-head attention network combining explicit and implicit information

In event-based social networks (EBSN), group event recommendation has become an important task for groups to quickly find events that they are interested in. Existing methods on group event recommendation either consider just one type of information, explicit or implicit, or separately model the explicit and implicit information. However, these methods often generate a problem of data sparsity or of model vector redundancy. In this paper, we present a Graph Multi-head Attention Network (GMAN) model for group event recommendation that integrates the explicit and implicit information in EBSN. Specifically, we first construct a user-explicit graph based on the user's explicit information, such as gender, age, occupation and the interactions between users and events. Then we build a user-implicit graph based on the user's implicit information, such as friend relationships. The incorporated both explicit and implicit information can effectively describe the user's interests and alleviate the data sparsity problem. Considering that there may be a correlation between the user's explicit and implicit information in EBSN, we take the user's explicit vector representation as the input of the implicit information aggregation when modeling with graph neural networks. This unified user modeling can solve the aforementioned problem of user model vector redundancy and is also suitable for event modeling. Furthermore, we utilize a multi-head attention network to learn richer implicit information vectors of users and events from multiple perspectives. Finally, in order to get a higher level of group vector representation, we use a vanilla attention mechanism to fuse different user vectors in the group. Through experimenting on two real-world Meetup datasets, we demonstrate that GMAN model consistently outperforms state-of-the-art methods on group event recommendation.     

BiRSwinT: Bilinear full-scale residual swin-transformer for fine-grained driver behavior recognition

Driver behavior recognition has attracted extensive attention recently. Numerous methods have been developed on the basis of various deep neural networks. However, the existing models still suffer from various challenges in solving the downstream fine-grained recognition task, because many driver behaviors share similar global shape / appearances, and can only be discriminated by subtle but meaningful actions of local body parts. To remedy this emerging issue of fine-grained driver behavior recognition, we present a bilinear full-scale residual Swin-Transformer network (BiRSwinT) to learn and fuse the global shape/appearance and the local discriminative cues of driver actions. Specifically, a transformer-based dual-stream structure is designed, which contains two parallel branches: a global representation branch and a local residual branch. To learn the multi-scale local cues implied in driver actions, a set of full-scale skip connections are introduced in the local residual branch. Then a bilinear fusion with multi-step training strategy is employed to leverage and combine the global representation and local detail features of driver actions. The proposed method has been validated on the AUC V1 and V2 datasets, and achieved the average Top-1 accuracy of 93.235% and 92.253%, respectively. Comparison experiments show that the proposed method has superior performance when compared to the state-of-the-art methods. The generalization experiments on the State-Farm dataset also demonstrate the superiority of the proposed framework and strategy. The source code is available at https://github.com/shadow2469/BiRSwinT.git.     

Heterogeneous graph-based joint representation learning for users and POIs in location-based social network

Learning latent representations for users and points of interests (POIs) is an important task in location-based social networks (LBSN), which could largely benefit multiple location-based services, such as POI recommendation and social link prediction. Many contextual factors, like geographical influence, user social relationship and temporal information, are available in LBSN and would be useful for this task. However, incorporating all these contextual factors for user and POI representation learning in LBSN remains challenging, due to their heterogeneous nature. Although the encouraging performance of POI recommendation and social link prediction are delivered, most of the existing representation learning methods for LBSN incorporate only one or two of these contextual factors. In this paper, we propose a novel joint representation learning framework for users and POIs in LBSN, named UP2VEC. In UP2VEC, we present a heterogeneous LBSN graph to incorporate all these aforementioned factors. Specifically, the transition probabilities between nodes inside the heterogeneous graph are derived by jointly considering these contextual factors. The latent representations of users and POIs are then learnt by matching the topological structure of the heterogeneous graph. For evaluating the effectiveness of UP2VEC, a series of experiments are conducted with two real-world datasets (Foursquare and Gowalla) in terms of POI recommendation and social link prediction. Experimental results demonstrate that the proposed UP2VEC significantly outperforms the existing state-of-the-art alternatives. Further experiment shows the superiority of UP2VEC in handling cold-start problem for POI recommendation.     

Phishing scams detection via temporal graph attention network in Ethereum

Recently, phishing scams have become one of the most serious types of crime involved in Ethereum, the second-largest blockchain-based cryptocurrency platform. The existing phishing scams detection techniques for Ethereum mostly use traditional machine learning or network representation learning to mine the key information from the transaction network and identify phishing addresses. However, these methods typically crop the temporal transaction graph into snapshot sequences or construct temporal random wanderings to model the dynamic evolution of the topology of the transaction graph. In this paper, we propose PDTGA, a method that applies graph representation learning based on temporal graphs attention to improve the effectiveness of phishing scams detection in Ethereum. Specifically, we learn the functional representation of time directly and model the time signal through the interactions between the time encoding function and node features, edge features, and the topology of the graph. We collected a real-world Ethereum phishing scam dataset, containing over 250,000 transaction records between more than 100,000 account addresses, and divided them into three datasets of different sizes. Through data analysis, we first summarized the periodic pattern of Ethereum phishing scam activities. Then we constructed 14 kinds of account node features and 3 kinds of transaction edge features. Experimental evaluations based on the above three datasets demonstrate that PDTGA with 94.78% AUC score and 88.76% recall score outperforms the state-of-the-art methods.     

Contrastive Graph Convolutional Networks with adaptive augmentation for text classification

Text classification is an important research topic in natural language processing (NLP), and Graph Neural Networks (GNNs) have recently been applied in this task. However, in existing graph-based models, text graphs constructed by rules are not real graph data and introduce massive noise. More importantly, for fixed corpus-level graph structure, these models cannot sufficiently exploit the labeled and unlabeled information of nodes. Meanwhile, contrastive learning has been developed as an effective method in graph domain to fully utilize the information of nodes. Therefore, we propose a new graph-based model for text classification named CGA2TC, which introduces contrastive learning with an adaptive augmentation strategy into obtaining more robust node representation. First, we explore word co-occurrence and document word relationships to construct a text graph. Then, we design an adaptive augmentation strategy for the text graph with noise to generate two contrastive views that effectively solve the noise problem and preserve essential structure. Specifically, we design noise-based and centrality-based augmentation strategies on the topological structure of text graph to disturb the unimportant connections and thus highlight the relatively important edges. As for the labeled nodes, we take the nodes with same label as multiple positive samples and assign them to anchor node, while we employ consistency training on unlabeled nodes to constrain model predictions. Finally, to reduce the resource consumption of contrastive learning, we adopt a random sample method to select some nodes to calculate contrastive loss. The experimental results on several benchmark datasets can demonstrate the effectiveness of CGA2TC on the text classification task.     

Exploiting context-awareness and multi-criteria decision making to improve items recommendation using a tripartite graph-based model

Integrating useful input information is essential to provide efficient recommendations to users. In this work, we focus on improving items ratings prediction by merging both multiple contexts and multiple criteria based research directions which were addressed separately in most existent literature. Throughout this article, Criteria refer to the items attributes, while Context denotes the circumstances in which the user uses an item. Our goal is to capture more fine grained preferences to improve items recommendation quality using users’ multiple criteria ratings under specific contextual situations. Therefore, we examine the recommenders’ data from the graph theory based perspective by representing three types of entities (users, contextual situations and criteria) as well as their relationships as a tripartite graph. Upon the assumption that contextually similar users tend to have similar interests for similar item criteria, we perform a high-order co-clustering on the tripartite graph for simultaneously partitioning the graph entities representing users in similar contextual situations and their evaluated item criteria. To predict cluster-based multi-criteria ratings, we introduce an improved rating prediction method that considers the dependency between users and their contextual situations, and also takes into account the correlation between criteria in the prediction process. The predicted multi-criteria ratings are finally aggregated into a single representative output corresponding to an overall item rating. To guide our investigation, we create a research hypothesis to provide insights about the tripartite graph partitioning and design clear and justified preliminary experiments including quantitative and qualitative analyzes to validate it. Further thorough experiments on the two available context-aware multi-criteria datasets, TripAdvisor and Educational, demonstrate that our proposal exhibits substantial improvements over alternative recommendations approaches.     

A new deep graph attention approach with influence and preference relationship reconstruction for rate prediction recommendation

Graph neural networks have been frequently applied in recommender systems due to their powerful representation abilities for irregular data. However, these methods still suffer from the difficulties such as the inflexible graph structure, sparse and highly imbalanced data, and relatively shallow networks, limiting rate prediction ability for recommendations. This paper presents a novel deep dynamic graph attention framework based on influence and preference relationship reconstruction (DGA-IPR) for recommender systems to learn optimal latent representations of users and items. The entire framework involves a user branch and an item branch. An influence-based dynamic graph attention (IDGA) module, a preference-based dynamic graph attention (PDGA) module, and an adaptive fine feature extraction (AFFE) module are respectively constructed for each branch. Concretely, the first two attention modules concentrate on reconstructing influence and preference relationship graphs, breaking imbalanced and fixed constraints of graph structures. Then a deep feature aggregation block and an adaptive feature fusion operation are built, improving the network depth and capturing potential high-order information expressions. Besides, AFFE is designed to acquire finer latent features for users and items. The DGA-IPR architecture is formed by integrating IDGA, PDGA, and AFFE for users and items, respectively. Experiments reveal the superiority of DGA-IPR over existing recommendation models.     

A fine-grained and multi-context-aware learning path recommendation model over knowledge graphs for online learning communities

Existing approaches to learning path recommendation for online learning communities mainly rely on the individual characteristics of users or the historical records of their learning processes, but pay less attention to the semantics of users’ postings and the context. To facilitate the knowledge understanding and personalized learning of users in online learning communities, it is necessary to conduct a fine-grained analysis of user data to capture their dynamical learning characteristics and potential knowledge levels, so as to recommend appropriate learning paths. In this paper, we propose a fine-grained and multi-context-aware learning path recommendation model for online learning communities based on a knowledge graph. First, we design a multidimensional knowledge graph to solve the problem of monotonous and incomplete entity information presentation of the single layer knowledge graph. Second, we use the topic preference features of users’ postings to determine the starting point of learning paths. We then strengthen the distant relationship of knowledge in the global context using the multidimensional knowledge graph when generating and recommending learning paths. Finally, we build a user background similarity matrix to establish user connections in the local context to recommend users with similar knowledge levels and learning preferences and synchronize their subsequent postings. Experiment results show that the proposed model can recommend appropriate learning paths for users, and the recommended similar users and postings are effective.     

Graph neural news recommendation with long-term and short-term interest modeling

With the information explosion of news articles, personalized news recommendation has become important for users to quickly find news that they are interested in. Existing methods on news recommendation mainly include collaborative filtering methods which rely on direct user-item interactions and content based methods which characterize the content of user reading history. Although these methods have achieved good performances, they still suffer from data sparse problem, since most of them fail to extensively exploit high-order structure information (similar users tend to read similar news articles) in news recommendation systems. In this paper, we propose to build a heterogeneous graph to explicitly model the interactions among users, news and latent topics. The incorporated topic information would help indicate a user’s interest and alleviate the sparsity of user-item interactions. Then we take advantage of graph neural networks to learn user and news representations that encode high-order structure information by propagating embeddings over the graph. The learned user embeddings with complete historic user clicks capture the users’ long-term interests. We also consider a user’s short-term interest using the recent reading history with an attention based LSTM model. Experimental results on real-world datasets show that our proposed model significantly outperforms state-of-the-art methods on news recommendation.     

Combining Non-sampling and Self-attention for Sequential Recommendation

With the rapid development of social media and big data technology, user’s sequence behavior information can be well recorded and preserved on different media platforms. It is crucial to model the user preference through mining their sequential behaviors. The goal of sequential recommendation is to predict what a user may interact with in the next moment based on the user’s historical record of interactive sequence. However, existing sequential recommendation methods generally adopt a negative sampling mechanism (e.g. random and uniform sampling) for the pairwise learning, which brings the defect of insufficient training to the model, and decrease the evaluation performance of the entire model. Therefore, we propose a Non-sampling Self-attentive Sequential Recommendation (NSSR) model that combines non-sampling mechanism and self-attention mechanism. Under the premise of ensuring the efficient training of the model, NSSR model takes all pairs in the training set as training samples, so as to achieve the goal of fully training the model. Specifically, we take the interactive sequence as the current user representation, and propose a new loss function to implement the non-sampling training mechanism. Finally, the state-of-the-art result is achieved on three public datasets, Movielens-1M, Amazon Beauty and Foursquare_TKY, and the recommendation performance increase by about 29.3%, 25.7% and 42.1% respectively.     

HVLM: Exploring Human-Like Visual Cognition and Language-Memory Network for Visual Dialog

Visual dialog, a visual-language task, enables an AI agent to engage in conversation with humans grounded in a given image. To generate appropriate answers for a series of questions in the dialog, the agent is required to understand the comprehensive visual content of an image and the fine-grained textual context of the dialog. However, previous studies typically utilized the object-level visual feature to represent a whole image, which only focuses on the local perspective of an image but ignores the importance of the global information in an image. In this paper, we proposed a novel model Human-Like Visual Cognitive and Language-Memory Network for Visual Dialog (HVLM), to simulate global and local dual-perspective cognitions in the human visual system and understand an image comprehensively. HVLM consists of two key modules, Local-to-Global Graph Convolutional Visual Cognition (LG-GCVC) and Question-guided Language Topic Memory (T-Mem). Specifically, in the LG-GCVC module, we design a question-guided dual-perspective reasoning to jointly learn visual contents from both local and global perspectives through a simple spectral graph convolution network. Furthermore, in the T-Mem module, we design an iterative learning strategy to gradually enhance fine-grained textual context details via an attention mechanism. Experimental results demonstrate the superiority of our proposed model, which obtains the comparable performance on benchmark datasets VisDial v1.0 and VisDial v0.9.     

社会化标注系统中个性化信息推荐多维度融合与优化模型研究

[目的/意义]在社会化标注系统自组织运行的基础上，构建个性化信息推荐的多维度融合与优化模型，进而在大数据环境下，为用户提供精准的个性化信息推荐服务，从而进一步丰富个性化信息推荐的理论体系以及拓展个性化信息推荐的研究方法。[方法/过程]首先，对每一种个性化信息推荐方法的优点和不足进行深入分析；然后，将基于图论（社会网络关系）、基于协同过滤以及基于内容（主题）3种个性化信息推荐方法进行多维度深度融合，构建个性化信息推荐多维度融合模型；最后，对社会化标注系统中个性化信息推荐多维度融合模型进行优化，从而解决个性化推荐过程中用户"冷启动"、数据稀疏性和用户偏好漂移等问题。[结果/结论]通过综合考虑现有的基于图论（社会网络关系）、基于协同过滤以及基于内容（主题）的个性化信息推荐方法各自的贡献和不足，实现3种方法之间的多维度深度融合，并结合心理认知、用户情境以及时间、空间等优化因素，最终构建出社会化标注系统中个性化信息推荐多维度融合与优化模型。     

Dual disentanglement of user–item interaction for recommendation with causal embedding

To achieve personalized recommendations, the recommender system selects the items that users may like by learning the collected user–item interaction data. However, the acquisition and use of data usually form a feedback loop, which leads to recommender systems suffering from popularity bias. To solve this problem, we propose a novel dual disentanglement of user–item interaction for recommendation with causal embedding (DDCE). Different from the existing work, our innovation is we take into account double-end popularity bias from the user-side and the item-side. Firstly, we perform a causal analysis of the reasons for user–item interaction and obtain the causal embedding representation of each part according to the analysis results. Secondly, on the item-side, we consider the influence of item attributes on popularity to improve the reliability of the item popularity. Then, on the user-side, we consider the effect of the time series when obtaining users’ interest. We model the contrastive learning task to disentangle users’ long–short-term interests, which avoids the bias of long–short-term interests overlapping, and use the attention mechanism to realize the dynamic integration of users’ long–short-term interests. Finally, we realize the disentanglement of user–item interaction reasons by decoupling user interest and item popularity. We experiment on two real-world datasets (Douban Movie and KuaiRec) to verify the significance of DDCE, the average improvement of DDCE in three evaluation metrics (NDCG, HR, and Recall) compared to the state-of-the-art model are 5.1106% and 4.1277% (MF as the backbone), 3.8256% and 3.2790% (LightGCN as the backbone), respectively.     

Item diversified recommendation based on influence diffusion

Recently, the high popularity of social networks accelerates the development of item recommendation. Integrating the influence diffusion of social networks in recommendation systems is a challenging task since topic distribution over users and items is latent and user topic interest may change over time. In this paper, we propose a dynamic generative model for item recommendation which captures the potential influence logs based on the community-level topic influence diffusion to infer the latent topic distribution over users and items. Our model enables tracking the time-varying distributions of topic interest and topic popularity over communities in social networks. A collapsed Gibbs sampling algorithm is proposed to train the model, and an improved diversification algorithm is proposed to obtain item diversified recommendation list. Extensive experiments are conducted to evaluate the effectiveness and efficiency of our method. The results validate our approach and show the superiority of our method compared with state-of-the-art diversified recommendation methods.     

Global information-aware argument mining based on a top-down multi-turn QA model

Argument mining (AM) aims to automatically generate a graph that represents the argument structure of a document. Most previous AM models only pay attention to a single argument component (AC) to classify the type of the AC or a pair of ACs to identify and classify the argumentative relation (AR) between the two ACs. These models ignore the impact of global argument structure of the documents, which is important, especially in some highly structured genres such as scientific papers, where the process of argumentation is relatively fixed. Inspired by this, we propose a novel two-stage model which leverages global structure information to support AM. The first stage uses a multi-turn question-answering model to incrementally generate an initial argumentative graph that identifies relations among ACs. At each turn, all ACs related to the query AC are generated simultaneously, such that the sibling global information between the answer ACs is considered. In addition, the partially constructed graph is used as global structure information to support the extension of the graph with additional ACs. After the whole initial graph structure has been determined, the second stage assigns semantic types to both the ACs and ARs among them, leveraging information from this initial graph as global structure information. We test the proposed methods on two scientific datasets (one is the AbstRCT dataset including 659 abstracts about cancer research and the other is the SciARG dataset that consists of 225 computer linguistic abstracts and 285 biomedical abstracts) and a student essay dataset PE with 402 essays. Our experiments show that our model improves the state-of-the-art performance on two scientific datasets for different AM subtasks, with average improvements of 1%, 2.41%, 1.1% for the ACC, ARI and ARC task respectively on the AbstRCT dataset, and 2.36%, 1.84%, 8.87% for the ACC, ARI and ARC task on the SciARG dataset. Our model also achieves comparative results on the PE datasets: 87.7% of F1 scores for the ACC task, 81.4% for the ARI task and 78.8% for the ARC task.