Wu Shu（吴 书）

I am currently an Associated Professor in the Center for Research on Intelligent Perception and Computing (CRIPAC) , National Laboratory of Pattern Recognition (NLPR), Institute of Automation, Chinese Academy of Sciences.

My research interests include data mining, big data, and network data analytics. I am a senior member of IEEE, ACM and CCF.

Email  /  UCAS Homepage  /  Github  /  Google Scholar  

I'm interested in data mining, recommendation systems, pervasive computing. Representative papers are highlighted.

We propose a personalized IN terest S ustainability modeling framework for sequential POI REcommendation, INSPIRE for brevity. Different from existing methods that directly recommend next POIs through users’ historical trajectories, our proposed INSPIRE focuses on users’ personalized interest sustainability.

We propose a novel Latent Relations Learning method for Temporal Knowledge Graph reasoning. It employs a Structural Encoder to obtain entity representations and a Latent Relations Learning module to mine and exploit intra- and inter-time latent relations. The extracted temporal representations are then used for entity prediction.

We propose a novel method from a counterfactual view, namely CLEVER. It trains a claim-evidence fusion model and a claim-only model independently and obtains the final unbiased prediction by subtracting the output of the claim-only model from the output of the claim-evidence fusion model.

We propose a novel MOdality DEcorrelating STable learning framework, MODEST for brevity, to learn users’ stable preference. This method aims to estimate a weight for each item, such that the features from different modalities in the weighted distribution are decorrelated.

We propose a missing label imputation approach to improve multi-task molecular property prediction, using a bipartite graph to model molecule-task co-occurrence relationships. A graph neural network is used for predicting missing edges on the graph, and reliable pseudo-labels are selected based on prediction result uncertainty.

We propose a novel method that utilizes a designed Hierarchical Relational Graph Neural Network to learn the long- and short-term representations for TKG reasoning, namely HGLS. We first transform the TKG into a global graph. Based on the built graph, we design a Hierarchical Relational Graph Neural Network.

We proposes a novel network to explicitly model the enterprise credit rating problem using DNNs and attention mechanisms, allowing for explainable enterprise credit ratings. Experiments conducted on real-world enterprise datasets show that the proposed approach achieves higher performance.

We provide a comprehensive review of deep graph generation, encompassing methods, application areas, problem formulation, state-of-the-art categorization, generation strategies, and future challenges and opportunities.

We propose a novel global attention module from two levels: channel level and node level. Specifically, we exploit second-order channel correlation to extract more discriminative representations.

We summarize the mainstream frameworks and techniques that can be properly relied on for developing AIs for complex human-computer gaming; raise the challenges or drawbacks of current techniques in the successful AIs; and try to point out future trends in human-computer gaming AIs.

We introduce MICRO (MIning with ContRastive mOdality fusion model), a framework that learns item relationships within modalities and enables multimodal fusion. It enhances collaborative filtering for accurate recommendations by leveraging modality-aware structure learning and contrastive techniques.

We propose a novel Temporal Meta-learning framework for TKG reasoning, which learns evolutionary metaknowledge from temporal meta-tasks to adaptively handle future data and entities with limited historical information. The framework incorporates a Gating Integration module for establishing flexible temporal correlations between meta-tasks.

We propose GraphDIVE to enhance GNN performance and explore the connection between topological structure and class imbalance. GraphDIVE learns multi-view graph representations and combines them with multi-view experts (classifiers) to capture the diverse intrinsic characteristics of graph topological structure.

We propose a novel framework for debiasing evidence-based fake news detection1 by causal intervention. Under this framework, the model is first trained on the original biased dataset like ordinary work.

This work presents a novel Contrastive leaRning framEwork for Multiview network Embedding (CREME). In our work, different views can be obtained based on the various relations among nodes.

We propose DIGNN, a simple and effective method that leverages attention mechanism to adaptively fuse two views based on data-specific preference. Additionally, we enhance DIGNN by incorporating mutual information constraints for both topology and attribute, utilizing variational bounds to approximate the optimization objective function.

We aims at leveraging most existing graph embedding works into one formula via introducing the graph consensus term and proposes a unified and scalable multiview learning framework, termed graph consensus multiview framework (GCMF).

We propose a novel Reject-aware Multi-Task Network (RMT-Net), which learns the task weights that control the information sharing from the rejection/approval task to the default/non-default task by a gating network based on rejection probabilities.

We propose an efficient dynamic graph embedding method that extends GCN-based approaches. It efficiently updates node embeddings by propagating changes in topological structure and neighborhood embeddings. The update process prioritizes the most affected nodes, which then propagate their changes to neighboring nodes for further updates.

We propose a unified Graph-based sEmantic sTructure mining framework, namely GET in short. Specifically, different from the existing work that treats claims and evidences as sequences, we model them as graph-structured data and capture the long-distance semantic dependency among dispersed relevant snippets via neighborhood propagation.

We propose a novel multiview contrastiveaggregation objective to adaptively distill information from eachview. In addition, we advocate the explicit use of structure embed-ding, which enriches the model with local structural patterns of theunderlying HGs, so as to better mine true and hard negatives forGCL.

We propose a new method named Dynamic Graph Neural Network for Sequential Recommendation (DGSR), which connects different user sequences through a dynamic graph structure, exploring the interactive behavior of users and items with time and order information,

We propose a novel method, named Personalized Graph Neural Networks with Attention Mechanism (A-PGNN) for brevity.

We propose a general contrastive paradigm which characterizes previous work by limiting the design space of interest to four dimensions: (a) data augmentation functions, (b) contrasting modes, (c) contrastive objectives, and (d) negative mining strategies.

We propose a novel relevance matching model based on graph neural networks to leverage the documentlevel word relationships for ad-hoc retrieval.

We propose a Meta-learning-based Cold-Start Sequential Recommendation Framework, namely Mecos, to mitigate the item cold-start problem in sequential recommendation.

We propose a Graph-based Hierarchical Relevance Matching model (GHRM) for ad-hoc retrieval, by which we can capture the subtle and general hierarchical matching signals simultaneously.

We propose a fine-grained Disentangled Item Representation (DIR) for recommender systems in this paper, where the items are represented as several separated attribute vectors instead of a single latent vector.

We propose a LATent sTructure mining method for multImodal reCommEndation, which we term LATTICE for brevity.

We propose a novel DisentanglEd Self-atTentIve NEtwork (DESTINE) framework for CTR prediction that explicitly decouples the computation of unary feature importance from pairwise interaction.

We propose a novel Active Learning with DivErse iNterpretations (ALDEN) approach.

We propose a novel label-informed graph structure learning framework which incorporates label information explicitly through a class transition matrix.

We propose a fully hyperbolic GCN model for recommendation, where all operations are performed in hyperbolic space.

We propose a novel model called Motifaware Sequential Recommendation (MoSeR), which captures the motifs hidden in behavior sequences to model the micro-structure features.

We propose a novel graph contrastive representation learning method with adaptive augmentation that incorporates various priors for topological and semantic aspects of the graph.

We propose to leverage the relation-aware heterogeneous graph method for user profiling, which also allows capturing significant meta relations.

We propose a novel Independence Promoted Graph Disentangled Networks (IPGDN) to learn disentangled node representation while enhancing the independence among node representations.

We propose TextING for inductive text classification via GNN.

We propose a novel target attentive graph neural network (TAGNN) model for session-based recommendation.

We propose a novel Tensor-based Feature interaction Network (TFNet) model, which introduces an operating tensor to elaborate feature interactions via multi-slice matrices in multiple semantic spaces.

We propose a novel Interaction Machine (IM) model. IM is an ecient and exact implementation of high-order FM, whose time complexity linearly grows with the order of interactions and the number of feature elds.

We propose a hybrid scheme for generating graph views on both structure and attribute levels.

We propose Dynamic Graph Collaborative Filtering (DGCF), a novel framework leveraging dynamic graphs to capture collaborative and sequential relations of both items and users at the same time.

We present a novel GraphAIR framework which models the neighborhood interaction in addition to neighborhood aggregation.

We propose a Multi-View Rrecurrent Neural Network (MV-RNN) mode.

We propose to represent the multi-field features in a graph structure intuitively, where each node corresponds to a feature field and different fields can interact through edges.

We propose a CNN & Attention-based Sequential Feature Extractor (CASFE) module to capture the possible features of user behaviors at different time intervals.

We propose a novel deep Hierarchical Graph Convolutional Network (H-GCN) for semisupervised node classification.

We propose a Hierarchical Contextual Attention-based GRU (HCA-GRU) network.

We propose to optimize the cluster indicator, which representing the class labels is an intuitive reflection of the clustering structure.

We propose a semi-supervised method to learn the compatibility across categories.

We propose an Event2vec module to learn distributed representations of events in social media.

We propose Node-wise Graph Neural Networks (NGNN) which can better model node interactions and learn better node representations.

We propose to acquire the spatial preference by modeling distances between successive POIs.

This paper proposes to mimic different views as different relations in a knowledge graph for unified and view-specific embedding learning.

We propose an attention-based approach for identification of misinformation (AIM).

We propose a novel method, Convolutional Approach for Misinformation Identification (CAMI) based on Convolutional Neural Network (CNN).

we propose a DeepStyle method for learning style features of items and sensing preferences of users.

We propose a Recurrent Log-BiLinear (RLBL) model. It can model multiple types of behaviors in historical sequences with behavior-specific transition matrices.

We propose a novel subspace learning framework for incomplete and unlabeled multi-view data.

We propose a novel model named recurrent models with contextual layer, which can model the sequential measurement data and contextual data simultaneously to predict the trend of users’ BP.

We represent each context value with a latent vector, and model the contextual information as a semantic operation on the user and item.

In this study, a coupled topic model (CoTM) for recommendation with UGC is developed.

We propose a novel model, named Context-Aware Recurrent Neural Networks (CA-RNN).

This work proposes a novel personalized ranking model which incorporates implicit feedback with content information by making use of Factorization Machines.

We propose a novel model, Dynamic REcurrent bAsket Model (DREAM), based on Recurrent Neural Network (RNN).

We establish a Network Information Credibility Evaluation (NICE) platform, which collects a database of rumors that have been verified on Sina Weibo and automatically evaluates the information which is generated by users on social media but has not been verified.

We extend RNN and propose a novel method called Spatial Temporal Recurrent Neural Networks (ST-RNN).

In this paper, a novel incomplete multi-view clustering method is therefore developed, which learns unified latent representations and projection matrices for the incomplete multi-view data.

We propose a Hierarchical Interaction Representation (HIR) model, which models the mutual action among different entities as a joint representation.

We propose a novel Social-Relational Topic Model (SRTM), which can alleviate the effect of topic-irrelevant links by analyzing relational users’ topics of each link.

We propose a novel model, Convolutional Click Prediction Model (CCPM), based on convolution neural network.

In this work, we present a unified framework, named Personalized Semantic Ranking (PSR).

We propose Contextual Operating Tensor (COT) model, which represents the common semantic effects of contexts as a contextual operating tensor and represents a context as a latent vector.

We propose two practical 1-parameter outlier detection methods, named ITB-SS and ITB-SP, which require no user-defined parameters for deciding whether an object is an outlier.

We improve the conventional rating-based objective function by using ranking constraints as the supplementary regularization to restrict the searching of predicted ratings in smaller and more likely ranges, and develop a novel method, called RankSVD++, based on the SVD++ model.

• Senior Program Committee (SPC) member, The AAAI Conference on Artificial Intelligence (AAAI), 2018-2022.


• Program Committee (TPC) member, The International Joint Conference on Artificial Intelligence (WWW), 2019-2022.


• Program Committee (TPC) member, The Web Conference (WWW), 2019-2022.


• Program Committee (TPC) member, The ACM International WSDM Conference (WSDM), 2019-2022.


• Technical Program Committee (TPC) member, The 22nd Pacific-Asia Conference on Knowledge Discovery and Data Mining(PAKDD), 2018.


• Senior Member, China Computer Federation (CCF), 2017-.


• Associate Editor, Frontiers of Computer Science, 2016-.


• Co-Chair, Multimedia and Social Networking, The 28th IEEE International Conference on Advanced Information Networking and Applications (AINA-2014).

• Research on Fake News Detection Based on Evidence Reasoning and Propagation Modeling; 2024.01-2027.12; National Science Foundation of China (NSFC).


• Co-principal InvestigatorKey technologies and applications for cross-scale systematic learning of social big data; 2022.01-2025.12; National Science Foundation of China (NSFC).


• The Theory and Method for Detecting and Recognizing Fake Media Content in Social Network; 2020.01-2023.12; National Science Foundation of China (NSFC).


• The Precision Service Technology of scientific and technological big data for the Individual Needs of Classified Users;2019.07-2022.06;National Key Research and Development Program of China.


• Principal Investigator; Research on User Behavior Modeling Methods based on Fusion of Entity Feature and Sequential Information; 2018.01 - 2021.12; National Science Foundation of China (NSFC).


• Principal Investigator; Social Recommendation with Contextual Information of Entity and Interaction; 2015.01 - 2017.12; National Science Foundation of China (NSFC).


• Principal Investigator; Research on Modeling User Behavior Methods based on Fusion of Entity Feature and Sequence Analysis;2018.01 - 2020.12; Beijing Natural Science Foundation (BJNSF).


• Principal Investigator; CASIA-JD Finance, Intelligence Financial Risk Joint Laboratory.


• Principal Investigator; Research on recommendation algorithms for mobile games; 2017.05 - 2017.10; Tencent.


• Principal Investigator; Abnormal behavior detection based on deep neural network and sparse coding; 2016.09 - 2017.10; CCF-Venustech Hongyan Research Fund.


• Principal Investigator; Click through rate prediction based on recurrent neural network; 2016.08 - 2017.10; CCF-Tencent Open Fund.


• Principal Investigator; Collaborative Prediction of High Blood Pressure with Contextual Information; 2016.01 - 2016.06; iHealth and Andon.


• Principal Investigator; Public Opinion Monitoring System based on Multi-modal Data; 2015.01 - 2017.12; MSR-CNIC Windows Azure Project.

This awesome template is from here
备案编号：闽ICP备15019970号