Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Zhongzhi ShiSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works360 from public data
- Heterogeneous Graph Attention Network3,463
Extensive experimental results on three real-world heterogeneous graphs not only show the superior performance of the proposed model over the state-of-the-arts, but also demonstrate its potentially good interpretability for graph analysis.
- A survey of heterogeneous information network analysis1,112
A survey of heterogeneous information network analysis is provided, which introduces basic concepts of heterogeneous information network analysis, examines its developments on different data mining tasks, discusses some advanced topics, and point out some future research directions.
- Beyond Low-frequency Information in Graph Convolutional Networks868
An experimental investigation assessing the roles of low-frequency and high-frequency signals is presented, and a novel Frequency Adaptation Graph Convolutional Networks (FAGCN) with a self-gating mechanism is proposed, which can adaptively integrate different signals in the process of message passing.
- Leveraging Meta-path based Context for Top- N Recommendation with A Neural Co-Attention Model716
A novel deep neural network with the co-attention mechanism for leveraging rich meta-path based context for top-N recommendation and performs well in the cold-start scenario and has potentially good interpretability for the recommendation results.
- Self-supervised Heterogeneous Graph Neural Network with Co-contrastive Learning572
This paper proposes a novel co-contrastive learning mechanism for HGNNs, named HeCo, which differs from traditional contrastive learning which only focuses on contrasting positive and negative samples, and employs cross-view contrastive mechanism.
- A Survey on Heterogeneous Graph Embedding: Methods, Techniques, Applications and Sources480
This survey presents several widely deployed systems that have demonstrated the success of HG embedding techniques in resolving real-world application problems with broader impacts and summarizes the open-source code, existing graph learning platforms and benchmark datasets.
- Metapath-guided Heterogeneous Graph Neural Network for Intent Recommendation394
A metapath-guided heterogeneous Graph Neural Network to learn the embeddings of objects in intent recommendation as a Heterogeneous Information Network is proposed and Offline experiments on real large-scale data show the superior performance of the proposed MEIRec, compared to representative methods.
- Heterogeneous Graph Attention Networks for Semi-supervised Short Text Classification367
A novel heterogeneous graph neural network based method for semi-supervised short text classification, leveraging full advantage of few labeled data and large unlabeled data through information propagation along the graph is proposed.
- HeteSim: A General Framework for Relevance Measure in Heterogeneous Networks350
Empirical studies show that HeteSim can effectively and efficiently evaluate the relatedness of heterogeneous objects, which is crucial to many data mining tasks.
- Heterogeneous Graph Structure Learning for Graph Neural Networks334
This work makes the first attempt towards learning an optimal heterogeneous graph structure for HGNNs and proposes a novel framework HGSL, which jointly performs Heterogeneous Graph Structure Learning and GNN parameters learning for classification task.
- Meta-learning on Heterogeneous Information Networks for Cold-start Recommendation313
This work proposes a novel semantic-enhanced tasks constructor and a co-adaptation meta-learner to address the two questions for how to capture HIN-based semantics in the meta-learning setting, and how to learn the general knowledge that can be easily adapted to multifaceted semantics.
- Semantic Path based Personalized Recommendation on Weighted Heterogeneous Information Networks290
This paper is the first to propose the weighted HIN and weighted meta path concepts to subtly depict the path semantics through distinguishing different link attribute values, and proposes a semantic path based personalized recommendation method SemRec to predict the rating scores of users on items.
- Compare to The Knowledge: Graph Neural Fake News Detection with External Knowledge266
A novel end-to-end graph neural model called CompareNet is proposed, which compares the news to the knowledge base (KB) through entities for fake news detection and significantly outperforms state-of-the-art methods.
- One2Multi Graph Autoencoder for Multi-view Graph Clustering264
This paper makes the first attempt to employ deep learning technique for attributed multi-view graph clustering, and proposes a novel task-guided One2Multi graph autoencoder clustering framework that can jointly optimize the cluster label assignments and embeddings suitable forgraph clustering.
- Interpreting and Unifying Graph Neural Networks with An Optimization Framework244
A surprising connection is established between different propagation mechanisms with a unified optimization problem, showing that despite the proliferation of various GNNs, in fact, their proposed propagation mechanisms are the optimal solution optimizing a feature fitting function over a wide class of graph kernels with a graph regularization term.
- Adversarial Learning on Heterogeneous Information Networks231
A novel framework HeGAN for HIN embedding is developed, which trains both a discriminator and a generator in a minimax game and proposes a generalized generator, which samples "latent" nodes directly from a continuous distribution, not confined to the nodes in the original network as existing methods are.
- 221
- HGAT: Heterogeneous Graph Attention Networks for Semi-supervised Short Text Classification203
A novel heterogeneous graph neural network-based method for semi-supervised short text classification, leveraging full advantage of limited labeled data and large unlabeled data through information propagation along the graph.
- Hyperbolic Graph Attention Network185
This article exploits the graph attention network to learn robust node representations of graphs in hyperbolic spaces as the gyrovector space framework provides an elegant algebraic formalism forHyperbolic geometry, and utilizes this framework to learn the graph representations in hyperBolic spaces.
- Multi-objective community detection in complex networks185
It is demonstrated that a combination of two negatively correlated objectives under the multi-objective framework usually leads to remarkably better performance compared with either of the orignal single objectives, including even many popular algorithms.
- Specformer: Spectral Graph Neural Networks Meet Transformers176
This work introduces Specformer, which effectively encodes the set of all eigenvalues and performs self-attention in the spectral domain, leading to a learnable set-to-set spectral filter and design a decoder with learnable bases to enable non-local graph convolution.
- Learning to Pre-train Graph Neural Networks171
L2P-GNN is proposed, a self-supervised pre-training strategy for GNNs that attempts to learn how to fine-tune during the pre- training process in the form of transferable prior knowledge.
- Extract the Knowledge of Graph Neural Networks and Go Beyond it: An Effective Knowledge Distillation Framework170
This paper proposes a framework based on knowledge distillation that extracts the knowledge of an arbitrary learned GNN model (teacher model), and injects it into a well-designed student model, which is designed as a trainable combination of parameterized label propagation and feature transformation modules.
- Cash-Out User Detection Based on Attributed Heterogeneous Information Network with a Hierarchical Attention Mechanism158
This work proposes a novel hierarchical attention mechanism based cash-out user detection model, called HACUD, which enhances feature representation of objects through meta-path based neighbors exploiting different aspects of structure information in AHIN.
- Generalizing Graph Neural Networks on Out-of-Distribution Graphs154
This paper argues that the spurious correlation exists among subgraph-level units and analyzes the degeneration of GNN in causal view, and proposes a general causal representation framework for stable GNN, called StableGNN, which not only outperforms the state-of-the-arts but also provides a flexible framework to enhance existing GNNs.
- Graph Neural News Recommendation with Unsupervised Preference Disentanglement153
This paper model the user-news interactions as a bipartite graph and proposes a novel Graph Neural News Recommendation model with Unsupervised Preference Disentanglement, named GNUD, which can effectively improve the performance of news recommendation and outperform state-of-the-art news recommendation methods.
- Graph Foundation Models: Concepts, Opportunities and Challenges152
This article introduces the concept of Graph Foundation Models (GFMs), and offers an exhaustive explanation of their key characteristics and underlying technologies, and classify the existing work related to GFMs into three distinct categories, based on their dependence on graph neural networks and large language models.
- Network Schema Preserving Heterogeneous Information Network Embedding152
This paper makes the first attempt to study network schema preserving HIN embedding, and proposes a novel model named NSHE, which significantly outperforms the state-of-the-art methods.
- Graph Structure Estimation Neural Networks151
The proposed graph estimation neural networks GEN presents a structure model to fit the mechanism of GNNs by generating graphs with community structure, and an observation model that injects multifaceted observations into calculating the posterior distribution of graphs and is the first to incorporate multi-order neighborhood information.
- FlowScope: Spotting Money Laundering Based on Graphs151
Theoretical analysis shows that FlowScope provides guarantees in terms of the amount of money that fraudsters can transfer without being detected, and outperforms state-of-the-art baselines in accurately detecting the accounts involved in money laundering, in both injected and real-world data settings.
- Self-supervised Graph Neural Networks for Multi-behavior Recommendation142
This paper proposes a novel self-supervised graph collaborative filtering model for multi-behavior recommendation named S-MBRec, and proposes a star-style contrastive learning task to capture the embedding commonality between target and auxiliary behaviors, so as to alleviate the sparsity of supervision signal, reduce the redundancy among auxiliary behavior, and extract the most critical information.
- Temporal Network Embedding with Micro- and Macro-dynamics142
A novel temporal network embedding method with micro- and macro-dynamics, named $\rmM^2DNE $ is proposed, which significantly outperforms the state-of-the-arts not only in traditional tasks, e.g., network reconstruction, but also in temporal tendency-related tasks, such as scale prediction.
- CuCo: Graph Representation with Curriculum Contrastive Learning141
This paper introduces four graph augmentation techniques to obtain the positive and negative samples, and utilize graph neural networks to learn their representations, and proposes a novel curriculum contrastive learning framework for self-supervised graph-level representation, called CuCo.
- Relation Structure-Aware Heterogeneous Information Network Embedding140
This paper takes the structural characteristics of heterogeneous relations into consideration and proposes a novel Relation structure-aware Heterogeneous Information Network Embedding model (RHINE), which significantly outperforms the state-of-the-art methods in various tasks, including node clustering, link prediction, and node classification.
- Relevance search in heterogeneous networks139
Empirical studies show that HeteSim can effectively evaluate the relatedness of heterogeneous objects, and in the query and clustering tasks, it can achieve better performances than conventional measures.
- Hyperbolic Heterogeneous Information Network Embedding138
This paper proposes a novel hyperbolic heterogeneous information network embedding model that has superior performance on network reconstruction and link prediction tasks but also shows its ability of capture hierarchy structure in HIN via visualization.
- A link clustering based overlapping community detection algorithm138
A novel algorithm to discover overlapping communities based on link clustering, which applies genetic operation to cluster on links to discover groups of links that have the same characteristics.
- 136
- Multi-Component Graph Convolutional Collaborative Filtering130
This paper proposes a novel Multi-Component graph convolutional Collaborative Filtering (MCCF) approach to distinguish the latent purchasing motivations underneath the observed explicit user-item interactions, and empirical results on three real datasets and a synthetic dataset show the significant performance gains of MCCF.
- Lorentzian Graph Convolutional Networks129
This paper rebuilds the graph operations of hyperbolic GCNs with Lorentzian version, e.g., the feature transformation and non-linear activation, and proves some proposed graph operations are equivalent in different types ofhyperbolic geometry, which fundamentally indicates their correctness.
- GraphTranslator: Aligning Graph Model to Large Language Model for Open-ended Tasks128
This work proposes to bridge the pretrained GM and LLM by a Translator, named GraphTranslator, aiming to leverage GM to handle the pre-defined tasks effectively and utilize the extended interface of LLMs to offer various open-ended tasks for GM.
- 120
- Representation learning via Dual-Autoencoder for recommendation116
A new representation learning framework called Recommendation via Dual-Autoencoder (ReDa) is proposed, which simultaneously learns the new hidden representations of users and items using autoencoders, and develops a gradient descent method to learn hidden representations.
- Graph Representation Learning: Foundations, Methods, Applications and Systems113
This tutorial systematically review the foundations, techniques, applications and advances in graph representation learning, and first introduces the foundations on graph theory and graph Fourier analysis.
- Knowledge-Adaptive Contrastive Learning for Recommendation112
This paper proposes a novel algorithm named Knowledge-Adaptive Contrastive Learning (KACL), which can significantly improve the performance on top-K recommendation compared with state-of-the-art methods and introduces two learnable view generators to adaptively remove task-irrelevant edges during data augmentation.
- Key Player Identification in Underground Forums over Attributed Heterogeneous Information Network Embedding Framework112
In this paper, an intelligent system named iDetective is proposed and developed to automate the analysis of underground forums for the identification of key players (i.e., users who play the vital role in the value chain).
- Deep Collaborative Filtering with Multi-Aspect Information in Heterogeneous Networks106
A Neural network based Aspect-level Collaborative Filtering model (NeuACF) to exploit different aspect latent factors and NeuACf++ to fuse aspect-level latent factors with self-attention mechanism are proposed.
- Aspect-Level Deep Collaborative Filtering via Heterogeneous Information Networks86
A Neural network based Aspect-level Collaborative Filtering model (NeuACF) to exploit different aspect latent factors in recommender system as a heterogeneous information network and significantly outperforms both existing latent factor models and recent neural network models.
- Contrastive Pre-Training of GNNs on Heterogeneous Graphs79
This paper proposes a novel Contrastive Pre-Training strategy of GNNs on Heterogeneous Graphs (CPT-HG), to capture both the semantic and structural properties in a self-supervised manner, and designs semantic-aware pre-training tasks at both the relation- and subgraph-levels, and enhances their representativeness by employing contrastive learning.
- Federated Heterogeneous Graph Neural Network for Privacy-preserving Recommendation76
A federated heterogeneous graph neural network (FedHGNN) based framework, which facilitates collaborative training of a recommendation model using distributed HINs while protecting user privacy and introduces an HGNN model for recommendation, which conducts node- and semantic-level aggregations to capture recovered semantics.
- Pre-training on Large-Scale Heterogeneous Graph76
The proposed PT-HGNN designs both the node- and schema-level pre-training tasks to contrastively preserve heterogeneous semantic and structural properties as a form of transferable knowledge for various downstream tasks to relieve the label scarcity issues.
- 75
- Compact Graph Structure Learning via Mutual Information Compression75
This paper theoretically prove that if the authors optimize basic views and final view based on mutual information, and keep their performance on labels simultaneously, the final view will be a minimal sufficient structure, named CoGSL.
- 74
- 72
- A GENETIC ALGORITHM FOR DETECTING COMMUNITIES IN LARGE-SCALE COMPLEX NETWORKS70
A genetic algorithm with a special encoding schema for community detection in complex networks, which employs a metric, named modularity Q as the fitness function and applies a special locus-based adjacency encoding schema to represent the community partitions.
- Confidence May Cheat: Self-Training on Graph Neural Networks under Distribution Shift68
A novel Distribution Recovered Graph Self-Training framework (DR-GST), which could recover the distribution of the original labeled dataset and apply the loss correction method to improve the quality of pseudo labels.
- 66
- Characterizing super-spreading in microblog: An epidemic-based information propagation model66
A parameterized model, the SAIR model, based on well-known epidemic models is built to characterize super-spreading phenomenon in tweet information propagation accompanied with super-spreaders and shows that the SAIR model is much more promising than the conventional SIR model in characterizing a super-spreading event of information propagation.
- A Data-centric Framework to Endow Graph Neural Networks with Out-Of-Distribution Detection Ability65
This work designs a post-hoc framework with Adaptive Amplifier for Graph OOD Detection, named AAGOD, concentrating on data-centric manipulation, and designs a Learnable Amplifier Generator to customize amplifiers for different graphs, and proposes a Regularized Learning Strategy to train parameters with no OOD data required.
- Anomalous Trajectory Detection Using Recurrent Neural Network65
A method is proposed which characterizes the trajectory by learning the trajectory embedding which can capture the sequential information of the trajectory and depict the internal characteristics between anomalous and norm trajectory to address the potential data sparsity problem.
- Robust Heterogeneous Graph Neural Networks against Adversarial Attacks64
This work proposes a novel robust HGNN framework RoHe against topology adversarial attacks by equipping an attention purifier, which can prune malicious neighbors based on topology and feature, and introduces the metapath-based transiting probability as the prior criterion of the purifier.
- Geometric Disentangled Collaborative Filtering63
A novel generative GDCF model is proposed to learn geometric disentangled representations by inferring the high-level concepts associated with user intentions and various geometries, and the experimental results demonstrate the superiority of this model.
- SemRec: a personalized semantic recommendation method based on weighted heterogeneous information networks63
The weighted HIN and weighted meta path concepts are introduced to subtly depict the path semantics through distinguishing different link attribute values and a semantic path based personalized recommendation method SemRec is proposed to predict the rating scores of users on items.
- Your Style Your Identity: Leveraging Writing and Photography Styles for Drug Trafficker Identification in Darknet Markets over Attributed Heterogeneous Information Network62
This paper proposes and develops an intelligent system named uStyle-uID leveraging both writing and photography styles for drug trafficker identification at the first attempt and devise a learning model named vIdentifier to classify if a given pair of traffickers are the same individual.
- Few-shot Link Prediction in Dynamic Networks60
This paper proposes a novel model based on a meta-learning framework, dubbed as MetaDyGNN, for few-shot link prediction in dynamic networks and designs a simple and effective dynamic graph neural network (GNN) module to characterize the local structure of each node in meta- learning tasks.
- A New Genetic Algorithm for Community Detection56
A new genetic algorithm for community detection is proposed that uses the fundamental measure criterion modularity Q as the fitness function and a special locus-based adjacency encoding scheme is applied to represent the community partition.
- Debiased Graph Neural Networks With Agnostic Label Selection Bias54
A novel debiased GNNs (DGNN) with a differentiated decorrelation regularizer is proposed, which outperforms the state-of-the-art methods and DGNN is a flexible framework to enhance existing GNN's.
- Improving Distantly-Supervised Relation Extraction with Joint Label Embedding54
A novel multi-layer attention-based model to improve relation extraction with joint label embedding that makes full use of both structural information from Knowledge Graphs and textual information from entity descriptions to learn label embeddings through gating integration while avoiding the imposed noise with an attention mechanism.
- 51
- OpenHGNN: An Open Source Toolkit for Heterogeneous Graph Neural Network45
OpenHGNN defines a unified and standard pipeline for training and testing, which can allow users to run a model on a specific dataset with just one command line, and can be extended to meet users' customized needs.
- Topic-aware Heterogeneous Graph Neural Network for Link Prediction45
This paper proposes a Topic-aware Heterogeneous Graph Neural Network, named THGNN, to hierarchically mine topic-aware semantics for learning multi-facet node representations for link prediction in HGs and applies an alternating two-step aggregation mechanism including intra-metapath decomposition and inter-metAPath mergence.
- Can Large Language Models Improve the Adversarial Robustness of Graph Neural Networks?44
An LLM-based robust graph structure inference framework, LLM4RGNN is proposed, which distills the inference capabilities of GPT-4 into a local LLM for identifying malicious edges and an LM-based edge predictor for finding missing important edges, so as to recover a robust graph structure.
- Who You Would Like to Share With? A Study of Share Recommendation in Social E-commerce44
A heterogeneous graph neural network based share recommendation model, called HGSRec is proposed, which delicately designs a tripartite heterogeneous GNNs to describe the multifold characteristics of users and items, and dynamically fuses them via capturing potential ternary dependency with a dual co-attention mechanism.
- Ranking-based Clustering on General Heterogeneous Information Networks by Network Projection43
This paper studies the ranking-based clustering problem in a general heterogeneous information network and proposes a novel solution HeProjI, which can achieve better clustering and ranking performances compared to other well-established algorithms.
- Multi-label Ensemble Learning43
This paper designs two objective functions to evaluate the accuracy and diversity of multilabel base learners, respectively, and proposes a novel solution, called EnML, to effectively augment the accuracy as well as the diversity of multi-label base learners.
- A Survey on Spectral Graph Neural Networks42
It is shown that spectral GNNs can capture global information and have better expressiveness and interpretability, and is classified according to the spectrum information they use, \ie, eigenvalues or eigenvectors.
- Regularizing Graph Neural Networks via Consistency-Diversity Graph Augmentations42
This paper analyzes two representative semi-supervised learning algorithms: label propagation (LP) and consistency regularization (CR) and finds that LP utilizes the prior knowledge of graphs to improve consistency and CR adopts variable augmentations to promote diversity.
- Space4HGNN41
A unified framework covering most HGNNs is proposed, consisting of three components: heterogeneous linear transformation, heterogeneous graph transformation, and heterogeneous message passing layer, and a platform Space4HGNN is built, which offers modularized components, reproducible implementations, and standardized evaluation for HGNN's.
- GraphMSE: Efficient Meta-path Selection in Semantically Aligned Feature Space for Graph Neural Networks41
GraphMSE is proposed, an efficient heterogeneous GCN combined with automatic meta-path selection that outperforms state-of-the-art counterparts, figures out importantMeta-paths, and is dramatically (e.g. 200 times) more efficient.
- Local and Global Information Fusion for Top-N Recommendation in Heterogeneous Information Network41
This work proposes a unified model LGRec to fuse local and global information for top-N recommendation in HIN, which firstly model most informative local neighbor information for users and items respectively with a co-attention mechanism and learns effective relation representations between users and Items to capture rich information in Hin by optimizing a multi-label classification problem.
- Graph Mining for Cybersecurity: A Survey40
A comprehensive review of graph mining for cybersecurity, including an overview of cybersecurity tasks, the typical graph mining techniques, and the general process of applying them to cybersecurity, as well as various solutions for different cybersecurity tasks.
- Data-Centric Graph Learning: A Survey39
A novel taxonomy based on the stages in the graph learning pipeline is proposed, and the processing methods for different data structures in the graph data are highlighted, i.e., topology, feature and label.
- NeuCast: Seasonal Neural Forecast of Power Grid Time Series38
NeuCast is proposed, a seasonal neural forecasting method, which dynamically models various loads as co-evolving time series in a hidden space, as well as extra weather conditions, in a neural network structure.
- On selection of objective functions in multi-objective community detection38
The experiments show that NSGA-Net optimizing over a pair of negatively correlated objectives usually performs better than the single-objective algorithm optimizing over either of the original objectives, and even better than other well-established community detection approaches.
- 36
- Dynamic Heterogeneous Graph Embedding via Heterogeneous Hawkes Process35
A novel Heterogeneous Hawkes Process based dynamic Graph Embedding (HPGE) is proposed that effectively integrates the Hawkes process into graph embedding to capture the excitation of various historical events on the current type-wise events.
- 35
- Relevance Measure in Large-Scale Heterogeneous Networks35
A novel measure, called AvgSim, which can measure similarity of same or different-typed object pairs in a uniform framework and has some good properties, is proposed and implemented using MapReduce parallel model in order to enable the application in large-scale networks.
- 31
- Profiling the Design Space for Graph Neural Networks based Collaborative Filtering31
This work makes the first attempt to profile the design space of GNN-based CF methods to enrich the understanding of different design dimensions as well as provide a novel paradigm of model design.
- Constrained-meta-path-based ranking in heterogeneous information network31
This paper proposes a constrained meta path to subtly capture the refined semantics through confining constraints on objects and studies the ranking problem in heterogeneous networks and proposes the HRank method to evaluate the importance of multiple types of objects and meta paths.
- Personalized Ranking Recommendation via Integrating Multiple Feedbacks30
A unified and flexible personalized ranking framework MFPR is proposed to integrate multiple feedbacks, and an explicit feedback based training data generation algorithm is designed to generate item pairs with more accurate partial order consistent with themultiple feedbacks for the proposed ranking model.
- Hierarchical Contrastive Learning Enhanced Heterogeneous Graph Neural Network29
HeCo++ conducts hierarchical contrastive learning, including cross-view and intra-view contrasts, which aims to enhance the mining of respective structures and shows superior performance over the state-of-the-arts.
- Co-clustering Interactions via Attentive Hypergraph Neural Network29
An attentive hypergraph neural network to encode the entire interactions, where an attention mechanism is utilized to select important attributes for explanations, and a novel co-clustering method to perform a joint clustering for the representations of interactions and the corresponding distributions of attribute selection, namely cluster-based consistency.
- COMPARISON AND SELECTION OF OBJECTIVE FUNCTIONS IN MULTIOBJECTIVE COMMUNITY DETECTION29
Extensive experiments on both artificial and real networks demonstrate that NSGA‐Net optimizing over a pair of negatively correlated objectives usually leads to better performances compared with the single‐objective algorithm optimizing over either of the original objectives, or even to other well‐established community detection approaches.
- A fast multi-objective evolutionary algorithm based on a tree structure29
The evolutionary algorithm based on dominating tree (DTEA) integrates the convergence strategy and diversity strategy into the DT and employs a DT-based eliminating strategy that realizes elitism and preserves population diversity without extra time and space costs.
- 28
- 28
- 28
- Neural Information Diffusion Prediction with Topic-Aware Attention Network27
This paper proposes Topic-aware Attention Network (TAN) to take advantage of both topic-specific diffusion modeling and deep learning techniques, and jointly model the text content of information items and cascade sequences by incorporating topical context and user/position dependencies into user representations via attention mechanisms.
- Heterogeneous Graph Neural Networks26
This chapter will give a brief review of the recent development on HG embedding, then introduce typical methods from the perspective of shallow and deep models, especially HGNNs, and point out future research directions for HGnns.
- Serum NLRP3 Inflammasome and BDNF: Potential Biomarkers Differentiating Reactive and Endogenous Depression26
The findings suggested the serum NLRP3 and BDNF levels could be potential biomarkers for detecting and evaluating the severity of reactive depression.
- Graph neural entity disambiguation26
A novel end-to-end graph neural entity disambiguation model which fully exploits the global semantic information is proposed which is applied on the entity-word graph to generate enhanced entity embeddings encoding global semantics.
- 26
- Sequence-aware Heterogeneous Graph Neural Collaborative Filtering25
A novel Sequence-aware Heterogeneous graph neural Collaborative Filtering model, called SHCF, is proposed, which can address the above problems by considering both the high-order heterogeneous collaborative signals and sequential information.
- Playlist Recommendation Based on Reinforcement Learning25
This paper proposes a novel reinforcement learning based model, called RLWRec, to exploit the optimal strategy of playlist, and proposes two novel strategies designed to solve the curse of state space and efficient music recommendation.
- A fast distributed stochastic Gradient Descent algorithm for matrix factorization25
This paper proposes a fast distributed SGD algorithm named FDSGD for matrix factorization, which can run efficiently in DCE and solves data sharing problem based on independent storage system to avoid data synchronization, and synchronous operation problem in D CE using a distributed synchronization tool so that distributed cooperation threads can perform in a harmonious environment.
- 24
- Social Influence Attentive Neural Network for Friend-Enhanced Recommendation24
In order to fuse rich heterogeneous information, the attentive feature aggregator in SIAN is designed to learn user and item representations at both node and typelevels, and a social influence coupler is put forward to capture the influence of the friend referral circle in an attentive manner.
- Entity set expansion in knowledge graph: a heterogeneous information network perspective24
An extended model CoMeSE++ is proposed which combines both structural information revealed by a KG and text information in Wikipedia for ESE and achieves better performance by combining structural and textual information of entities.
- A Novel Genetic Algorithm for Overlapping Community Detection24
A novel algorithm to discover overlapping communities based on edge clustering is proposed, which is different from conventional algorithms based on node clustering and will discover groups of edges that have the same characteristics.
- 24
- 23
- 23
- Coupled matrix factorization and topic modeling for aspect mining22
A novel model to effectively integrate Matrix factorization and Topic modeling for Aspect rating prediction (MaToAsp) is proposed, which achieves the best aspect rating prediction performance, compared to recent representative baselines.
- Abnormal Event Detection via Heterogeneous Information Network Embedding22
A novel deep heterogeneous network embedding method which incorporates the entity attributes and second-order structures simultaneously to address the problem of number of co-occurrences of majority entities in events.
- 22
- Graph Fairness Learning under Distribution Shifts21
To guarantee fairness performance on unknown testing graphs, a graph generator is proposed to produce numerous graphs with significant bias and under different distributions to achieve high classification and fairness performance even on generated graphs with significant bias, thereby effectively handling unknown testing graphs.
- 21
- Temporal Heterogeneous Interaction Graph Embedding for Next-Item Recommendation21
THIG Embedding (THIGE) is proposed which models the complex dynamics so that evolving short-term demands are guided by long-term historical habits, and leverages the rich heterogeneity to express the latent relevance of different-typed preferences.
- Spatial-Temporal Recurrent Neural Network for Anomalous Trajectories Detection21
An efficient method named Spatial-Temporal Recurrent Neural Network (ST-RNN) using coordinate sequence and spatial-temporal sequence is proposed that achieves state-of-the-art performance in most cases.
- A Two-Phase Model for Retweet Number Prediction21
A two-phase model to predict how many times a tweet can be retweeted in Sina Weibo is proposed, which first classifies tweets into several categories, and then does regression on each category.
- Multi-Label Classification Based on Multi-Objective Optimization21
This article proposes a novel solution to optimize over multiple objectives simultaneously of multi-objective multi-label classification and finds a set of non-dominated solutions which are optimal according to different trade-offs among multiple objectives.
- Collapsed Gibbs sampling for latent Dirichlet allocation on spark21
A collapsed Gibbs sampling method for the widely used latent Dirichlet allocation (LDA) model on Spark, which reduces data communication overhead, makes good use of Spark's efficient iterative execution and results in significant speedup on large-scale datasets in the authors' experiments.
- HRank: A Path Based Ranking Method in Heterogeneous Information Network21
The HRank method to evaluate the importance of multiple types of objects and meta paths and a constrained meta path is proposed to subtly capture the rich semantics in heterogeneous networks.
- Improving drug–drug interactions prediction with interpretability via meta-path-based information fusion20
The results of ablation study and cold-start scenario indicate that the meta-path-based information fusion mechanism red is beneficial for capturing the complex semantics among drug-related biological entities.
- 20
- Entity Set Expansion with Meta Path in Knowledge Graph20
This paper considers the knowledge graph as a heterogeneous information network (HIN) that contains different types of objects and links, and proposes a novel method, called MP_ESE, to extend entities in the HIN, which employs meta paths, a relation sequence connecting entities, in HIN to capture the implicit common traits of seed entities.
- Integrating Clustering and Ranking on Hybrid Heterogeneous Information Network20
Study of the ranking-based clustering on heterogeneous information network and proposed the ComClus algorithm, which applies star schema with self loop to organize the hybrid network and uses a probability model to represent the generative probability of objects.
- Artificial Intelligence for Complex Network: Potential, Methodology and Application19
This tutorial will explore the fascinating domain of empirical network modeling through artificial intelligence techniques, with applications across social media, web systems, and urban environments, to provide a comprehensive understanding of how these models can effectively represent, predict, and control empirical networked systems with heterogeneous structures and dynamic processes.
- Endowing Pre-trained Graph Models with Provable Fairness19
A novel adapter-tuning framework that endows pre-trained Graph models with Provable fAiRness (called GraphPAR), which freezes the parameters of PGMs and trains a parameter-efficient adapter to flexibly improve the fairness of PGMs in downstream tasks.
- 19
- FLAG: Fraud Detection with LLM-enhanced Graph Neural Network18
This paper proposes a novel framework, Fraud Detection with LLM-enhanced Graph Neural Networks (FLAG), and designs a fine-tuning approach that enables the LLM to extract discriminative text more closely related to the node labels, enhancing the model's ability to differentiate between fraudulent and normal nodes.
- PolarGate: Breaking the Functionality Representation Bottleneck of And-Inverter Graph Neural Network18
This work focuses on breaking the bottleneck of AIGNNs by augmenting their capability of functional representation, providing an efficient solution called PolarGate, which naturally aligns the message passing process with the logical functionality of AIGs and design a functionality-aware message passing strategy.
- Learning discrete adaptive receptive fields for graph convolutional networks18
A novel framework of conducting graph convolutions, where nodes are discretely selected among multi-hop neighborhoods to construct adaptive receptive fields (ARFs), which enable GCNs to get rid of the smoothness of soft attention weights, as well as to efficiently explore long-distance dependencies in graphs.
- 18
- 17
- Graph Distillation with Eigenbasis Matching17
Eigenbasis matching for spectrum-free graph condensation is proposed, named GCEM, which has two key steps: First, GCEM matches the eigenbasis of the real and synthetic graphs, rather than the graph structure, which eliminates the spectrum bias of GNNs.
- 17
- Accelerating Large-Scale Heterogeneous Interaction Graph Embedding Learning via Importance Sampling17
In real-world problems, heterogeneous entities are often related to each other through multiple interactions, forming a Heterogeneous Interaction Graph (HIG), and while modeling HIGs to deal with funda...
- Attention Based Meta Path Fusion for Heterogeneous Information Network Embedding17
In this paper, a novel attention based meta path fusion model for HIN embedding (called AMPE) is proposed, which extracts multiple homogeneous networks from HIN with meta paths, and then employs adopted AutoEncoders to embed theseHomogeneous networks.
- Multi-Objective Multi-Label Classification17
The Moml algorithm finds a set of non-dominated solutions which are optimal according to the different tradeoffs of the multiple objectives, which helps to provide more meaningful classification results in different application scenarios.
- A dominance tree and its application in evolutionary multi-objective optimization17
A data structure for storing the whole population and their dominating information in MOEAs, called a Dominance Tree (DT), is a binary tree that can effectively and efficiently store three-valued relations (namely dominating, dominated or non-dominated) among vector values.
- 16
- Text-Graph Enhanced Knowledge Graph Representation Learning16
This paper proposes to model the whole auxiliary text corpus with a graph and presents an end-to-end text-graph enhanced KG embedding model, named Teger, which significantly outperforms several state-of-the-art methods.
- FolkDiffusion: A Graph-Based Tag Suggestion Method for Folksonomies16
A heat diffusion method, i.e., FolkDiffusion, to rank tags for tag suggestions, which can suggest user- and resource-specific tags and prevent from topic drift is presented.
- Graph foundation model15
The concept of GFMs is explained and the key technologies in building GFMs are identified as the pre-train and adaptation techniques from the fields of GNNs and LLMs, to highlight their similarities and differences.
- Calibrating Graph Neural Networks from a Data-centric Perspective15
It is argued that the miscalibration of GNNs may stem from the graph data and can be alleviated through topology modification and proposed Data-centric Graph Calibration (DCGC), which uses two edge weighting modules to adjust the input graph for GNN calibration.
- 15
- Package Recommendation with Intra- and Inter-Package Attention Networks15
A first study on package recommendation is made and an Intra- and inter-package attention network for Package Recommendation (IPRec) is proposed, where an intra- package attention network is put forward to capture the object-level intention of user interacting with the package.
- Decorrelated Clustering with Data Selection Bias15
A novel Decorrelation regularized K-Means algorithm for clustering with data selection bias that aims to learn the global sample weights which are capable of balancing the sample distribution, so as to remove unexpected correlations among features.
- Minimum Topology Attacks for Graph Neural Networks14
An attack model, named MiBTack, based on a dynamic projected gradient descent algorithm, which can effectively solve the involving non-convex constraint optimization on discrete topology and can be used to measure node robustness, which is beyond what the current fixed-budget topology attacks can offer.
- Learning to Distill Graph Neural Networks14
This work proposes a general knowledge distillation framework that can be applied on any pretrained GNN models to further improve their performance, and introduces a scalable variant of the method to accelerate model training.
- Large-scale Comb-K Recommendation14
The comb-K recommendation model is proposed, a constrained combinatorial optimization model which seamlessly integrates the selection phase and delivery phase with delicately designed constraints and is able to simultaneously search the optimal combination of item selection and delivery with the full consideration of all user preferences.
- 14
- Multiplex Memory Network for Collaborative Filtering14
Inspired by recent progress in memory networks, a novel multiplex memory network for collaborative recommender systems (MMCF) is proposed, which leverages a multiplex memory layer consisting of an interaction memory and two co-occurrence context memories simultaneously, in order to jointly capture and locate important and relevant information in both user-item interactions and co-occurrence contexts.
- Local Low-Rank Matrix Approximation with Preference Selection of Anchor Points14
A new model for local low-rank matrix approximation which selects anchor-points using a heuristic method is proposed which outperforms many state-of-the-art recommendation methods, can significantly improve algorithm efficiency, and is easy to parallelize.
- Stock Time Series Forecasting Using Support Vector Machines Employing Analyst Recommendations14
An improved method to evaluate if an upgrade classification of SVM is reliable is put forward, and some interesting hints are discussed to help readers understand this model more explicitly.
- Learning Node Representations from Noisy Graph Structures13
This paper designs two generators, namely a graph generator and a noise generator, to identify normal structures and noises in an unsupervised setting and proposes a novel framework to learn noise-free node representations and eliminate noises simultaneously.
- Efficient Multi-task Prompt Tuning for Recommendation12
A novel two-stage prompt-tuning MTL framework (MPT-Rec) is proposed to address task irrelevance and training efficiency problems in multi-task recommender systems and achieves the best performance compared to the SOTA multi-task learning method on three real-world datasets.
- 12
- Lecture-style Tutorial: Towards Graph Foundation Models11
This tutorial will introduce the concept of graph foundation models, and provide a comprehensive exposition on their key characteristics and underpinning technologies, and thoroughly review existing works that lay the groundwork towards GFMs, which are summarized into three primary categories based on their roots in graph neural networks, large language models, or a hybrid of both.
- 11
- 11
- Editorial: Traditional Chinese medicine for depression and anxiety11
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY).
- Embedding Heterogeneous Information Network in Hyperbolic Spaces11
This article proposes a novel HIN embedding model HHNE, which employs the meta-path guided random walk to capture the structure and semantic relations between nodes, and derives an effective optimization strategy to update the hyperbolic embeddings iteratively.
- 11
- 11
- Advancing Molecular Graph-Text Pre-training via Fine-grained Alignment10
FineMolTex, a novel Fine-grained Molecular graph-Text pre-training framework to jointly learn coarse-grained molecule-level knowledge and fine-grained motif-level knowledge, is proposed and successfully captures fine-grained knowledge.
- GraphTeam: Facilitating Large Language Model-based Graph Analysis via Multi-Agent Collaboration10
This work proposes a multi-agent system based on LLM-based agents named GraphTeam, for graph analysis, which achieves state-of-the-art performance with an average 25.85% improvement over the best baseline in terms of accuracy.
- Prohibited Item Detection via Risk Graph Structure Learning10
RGSL first introduces structure learning into large-scale risk graphs, to reduce noisy connections and add similar pairs, and designs the pairwise training mechanism, which transforms the detection process as a metric learning from candidates to their similar prohibited items.
- Prohibited Item Detection on Heterogeneous Risk Graphs10
HSPD first designs the heterogeneous self-supervised learning model, which treats multiple semantics as the supervision to enhance item representations, and integrates with self-training mechanisms to iteratively expand confident pseudo labels for enriching supervision.
- 10
- Semi-supervised Co-Clustering on Attributed Heterogeneous Information Networks10
This paper designs a learnable overall relevance measure that integrates the structural and attributed relevance by employing meta-paths and attribute projection and proposes a novel approach, called SCCAIN, to co-cluster heterogeneous nodes based on constrained orthogonal non-negative matrix tri-factorization.
- Autonomic discovery of subgoals in hierarchical reinforcement learning10
The unique-direction-value (UDV) approach is chosen to form options offline and online autonomically and experiments show that the approach can find subgoals correctly, so the Q-learning with options found on bothOffline and online process can accelerate learning significantly.
- 10
- 10
- Benchmarking Graph Foundation Models9
GFMBench is proposed, an open-source pipeline that standardizes the training, evaluation, and deployment of GFMs across diverse real-world graph applications, and provides a unified benchmark for GFMs, enabling deeper insights into their generalization ability across various graph tasks and domains.
- 9
- A Post-Training Framework for Improving Heterogeneous Graph Neural Networks9
This paper proposes a general post-training framework that can be applied on any pretrained HGNNs to further inject task-specific knowledge and enhance their prediction performance, and designs an auxiliary system that estimates node labels based on a global inference module of multi-channel label propagation and a local inference modules of network schema-aware prediction.
- A Heterogeneous Information Network Method for Entity Set Expansion in Knowledge Graph9
The Heterogeneous Information Network (HIN) is used to represent Knowledge Graph (KG) to represent KG, which would effectively capture hidden semantic relations between seed entities and can quickly and accurately detect important path features in KG.
- Abstractive Document Summarization via Bidirectional Decoder9
A Summarization model using a Bidirectional decoder (BiSum), in which the backward decoder provides a reference for the forward decoder, which can produce higher-quality summary on Chinese datasets TTNews and English datasets CNN/Daily Mail.
- 9
- CORONA: A Coarse-to-Fine Framework for Graph-based Recommendation with Large Language Models8
This work proposes to leverage LLMs' reasoning abilities during the candidate filtering process, and introduces Chain Of Retrieval ON grAphs (CORONA) to progressively narrow down the range of candidate items on interaction graphs to enhance overall recommendation performance.
- 8
- xTrimoDock: Rigid Protein Docking via Cross-Modal Representation Learning and Spectral Algorithm8
This work develops a method called, xTrimoDock, which leverages a cross-modal representation learning to predict the protein distance map, and then uses a spectral initialization and gradient descent to obtain the roto-translation transformation for docking.
- Datasets and Interfaces for Benchmarking Heterogeneous Graph Neural Networks8
This work presents two large-scale, real-world, and challenging heterogeneous graph datasets from real scenarios: risk commodity detection and takeout recommendation and establishes standard benchmark interfaces that provide over 40 heterogeneousGraph Neural Networks datasets.
- 8
- 8
- On Balancing Neighborhood and Global Replacement Strategies in MOEA/D8
This work proposes a mechanism for adapting neighborhood and global replacement and designs an offspring generation strategy to generate high-quality solutions in a multiobjective evolutionary algorithm based on decomposition.
- 8
- Toward Graph-Tokenizing Large Language Models with Reconstructive Graph Instruction Tuning7
The key idea is to reconstruct the graph information from the LLM's graph token outputs, explicitly incorporating graph supervision to constrain the alignment process, resulting in a reconstructive graph instruction tuning pipeline, RGLM.
- Graph Foundation Models for Recommendation: A Comprehensive Survey7
This survey provides a comprehensive overview of GFM-based RS technologies by introducing a clear taxonomy of current approaches, diving into methodological details, and highlighting key challenges and future directions.
- Embedding geographic information for anomalous trajectory detection7
A novel method called A nomalous T rajectory D etection using R ecurrent N eural N etwork (ATD − RNN) which characterizes the trajectory with the learned trajectory embedding is proposed which can capture the sequential information of the trajectory and depict the internal characteristics between abnormal and normal trajectories.
- Matrix Factorization Meets Social Network Embedding for Rating Prediction7
A new framework for which combines network embedding and probabilistic matrix factorization is proposed, which makes use of social relation features extracted from social networks, on top of which an additional layer that uncovers the social dimensions that explain the variation in people’s feedback is learned.
- MFPR7
This work proposes a unified and flexible model, named Multiple Feedback-based Personalized Ranking (MFPR), to make full use of multiple feedback, which uses a personalized ranking framework.
- 7
- A Quick Multi-Objective Evolutionary Algorithm Based on Dominating Tree7
By examining three performance metrics on six test problems, the new algorithm is found to be competitive with SPEA2 and NSGA-II in terms of converging to the true Pareto front and maintaining the diversity of the population, moreover, it is much faster than other two algorithms.
- 6
- 6
- Advancing Molecule Invariant Representation via Privileged Substructure Identification6
A novel framework named MILI is proposed that formalizes molecule invariant learning based on privileged substructure identification and introduces substructure invariance constraint, and theoretically establishes two criteria for environment splits conducive to molecule invariant learning.
- 6
- Stigmatized experience is associated with exacerbated pain perception in depressed patients6
It is found that patients with depression who experienced stigmatization based on mental illness reported higher pain catastrophizing and performed increased pain perception for noxious stimuli than those who did not.
- Self-Supervised Spatio-Temporal Graph Learning for Point-of-Interest Recommendation6
A new method named SSTGL is proposed to combine self-supervised learning and GNN-based POI recommendation for the first time, empowered with spatio-temporal-aware strategies in the data augmentation and pre-text task stages, respectively, so that it can provide high-quality supervision information by incorporating spatio/temporal prior knowledge.
- 6
- 6
- RecExp6
This paper organizes the objects and relations in a recommender system with a heterogeneous information network, which integrates more informations and contains rich semantics and employs a semantic meta path based personalized recommendation model.
- 6
- Using WordNet in Conceptual Query Expansion6
This paper proposed a new query expansion technique using the comprehensive thesaurus WordNet and its semantic relatedness measure modules, demonstrating a 7% precision improvement over retrieval methods not employing query expansion techniques.
- Graph Positional Autoencoders as Self-supervised Learners5
This work proposes Graph Positional Autoencoders (GraphPAE), which employs a dual-path architecture to reconstruct both node features and positions and achieves state-of-the-art performance and consistently outperforms the baselines by a large margin.
- Harnessing Language Model for Cross-Heterogeneity Graph Knowledge Transfer5
A novel Language Model-enhanced Cross-Heterogeneity learning model, namely LMCH, which first design a metapath-based corpus construction method to unify HG representations as languages and proposes an iterative training pipeline with the help of an extra Graph Neural Network predictor, enhanced by LM-GNN contrastive alignment at the end of each iteration.
- Towards Text-free Graph Foundation Models: Rethinking Multi-Domain Graph Contrastive Learning5
This paper proposes a novel multi-domain pre-training and cross-domain transfer framework, namely MDGCL, which designs a contrastive learning strategy to substantially recognize and capture domain differences, and introduces domain tokens to encode domain-level global information.
- PIRLLS: Pretraining with Imitation and RL Finetuning for Logic Synthesis5
PIRLLS is proposed, a two-stage learning framework for imitation learning on expert trajectories followed by reinforcement learning (RL) finetuning, to efficiently explore the optimal synthesis flows and effectively improve the quality of logic optimization.
- Non-autoregressive personalized bundle generation5
This work proposes to perform the bundle generation via non-autoregressive mechanism and design a novel encoder-decoder framework named BundleNAT, which can effectively output the targeted bundle in one-shot without relying on any inherent order.
- Group-to-group recommendation with neural graph matching5
A group-to-group recommendation method, which abstracts user groups and item groups into graphs and calculates the similarity between two graphs based on graph matching, dubbed as GMRec, which achieves higher accuracy than state-of-the-art models on three industrial datasets with different group sizes.
- Customizing Graph Neural Network for CAD Assembly Recommendation5
This work proposes a novel neural architecture search (NAS) framework, dubbed CusGNN, which can design data-specific GNN automatically, and develops an effective differentiable search algorithm to search high-performing GNN from the search space.
- GammaGL: A Multi-Backend Library for Graph Neural Networks5
Gamma Graph Library (GammaGL), a GNN library that supports multiple deep learning frameworks as backends and uses a framework-agnostic design that allows users to easily switch between deep learning backends on top of existing components with a single line of code change is presented.
- 5
- 5
- Semantic-Specific Hierarchical Alignment Network for Heterogeneous Graph Adaptation5
This paper proposes a novel semantic-specific hierarchical alignment network for heterogeneous graph adaptation, called HGA, which designs a sharing-parameters HGNN ag-gregating path-based neighbors and hierarchical domain alignment strategies with the MMD and L 1 normalization term.
- Integrating Topic Model and Heterogeneous Information Network for Aspect Mining with Rating Bias5
A novel model integrating Topic model and HIN for Aspect Mining with rating bias is presented, designed to extract topic distributions of reviews by using textual information, and constructs a topic propagation network, and propagate topic distributions in this heterogeneous network.
- Author Set Identification via Quasi-Clique Discovery5
This paper proposes a novel Author Set Identification approach, namely ASI, which outperforms the state-of-art baselines in author set identification and leverages the learned embedding to construct a weighted paper-ego-network, which contains anonymous paper and candidate authors.
- A Hybrid Replacement Strategy for MOEA/D5
A hybrid mechanism balancing neighborhood and global replacement strategy, and an offspring generation method to generate the high-quality solution for each subproblem, which can ease mismatch between subproblems and solutions are proposed.
- Aspect Mining with Rating Bias5
Different from the widely used review-center models, RABI adopts the overall rating as the center of the probabilistic model, which generates reviews and topics, and a novel aspect rating variable in RABI is designed to effectively integrate the rating bias priori information.
- 5
- 5
- Data-centric Prompt Tuning for Dynamic Graphs4
DDGPrompt is proposed, a data-centric prompting framework designed to effectively refine pre-trained node embeddings at the input data level, enabling better adaptability to diverse downstream tasks.
- 4
- BurstEngine: An efficient distributed framework for training transformers On extremely Long sequences of over 1M tokens4
BurstEngine introduces BurstAttention, an optimized distributed attention with lower communication cost than RingAttention, an optimized distributed attention with lower communication cost than RingAttention that leverages topology-aware ring communication to fully utilize network bandwidth and incorporates fine-grained communication-computation overlap.
- MILS: Modality Interaction Driven Learning for Logic Synthesis4
MILS employs a modality interaction module to capture better and deeper relationships between circuit structures and synthesis sequences, enhancing the modeling capability of the synthesis process and outperforms the state-of-the-art (SOTA) methods.
- 4
- Retrieving GNN Architecture for Collaborative Filtering4
This work designs the neural retrieval approach based on meta-learning by developing two-level meta-features, ranking loss, and task-level data augmentation, and in a retrieval paradigm, RGCF can directly return a well-performing architecture given a new dataset (query), thus being efficient inherently.
- The Chinese Version of Oxford Depression Questionnaire: A Validation Study in Patients with Mood Disorders4
The reliability, structural validity, and criterion validity of the ODQ applied to patients with mood disorders meet the psychometric requirements, and the scale can be used to assess emotional blunting in Chinese patients with Mood disorders.
- Memory-Enhanced Period-Aware Graph Neural Network for General POI Recommendation4
A novel Memory-enhanced Period-aware Graph neural network for general POI Recommendation (MPGRec) exploits the advantages of the GNN module in characterizing user preferences and proposes a reading and writing strategy to merge the GNN module and memory module into a unified framework.
- A Joint Framework for Explainable Recommendation with Knowledge Reasoning and Graph Representation4
A joint framework for explainable recommendation with knowledge reasoning and graph representation is proposed, where a sub-graph is constructed from the paths generated through knowledge reasoning and utilized to optimize the KG representations.
- 4
- Recent Developments of Deep Heterogeneous Information Network Analysis4
This tutorial shall help researchers and practitioners to share new techniques for identifying and analyzing relationships in networks that integrate multiple types or sources of information.
- Multi-Objective Optimization for Overlapping Community Detection4
A Multi-Objective evolutionary algorithm for Overlapping Community Detection MOOCD is proposed and this algorithm simultaneously optimize two objective functions to get proper community partitions.
- 4
- An Improved Multiobjective Evolutionary Algorithm Based on Dominating Tree4
The nearest distance between the node and other nodes in the same sibling chain is used as its density estimation; the population growing and declining strategies are proposed to avoid the retreating and shrinking phenomenon caused by the restricted elite archive.
- Unifying Graph Neural Networks with a Generalized Optimization Framework3
This article establishes a surprising connection between different propagation mechanisms with an optimization problem, and shows that despite the proliferation of various GNNs, their proposed propagation mechanisms are the optimal solutions of a generalized optimization framework with a flexible feature fitting function and a generalized graph regularization term.
- PGAU: Static IR Drop Analysis for Power Grid using Attention U-Net Architecture and Label Distribution Smoothing3
This work designs a static IR drop analysis model for PG using the Attention U-Net architecture and proposes novel feature augmentation and selection methods to solve the feature confusion problem and use the label distribution smoothing (LDS) technique to handle unbalanced labels.
- Graph neural network based time estimator for SAT solver3
A novel model is proposed that integrates the working mechanism of SAT solvers with graph neural networks (GNNs) for predicting solving time, and incorporates a graph representation inspired by the CDCL paradigm that outperforms baselines in SAT solver runtime prediction.
- PMHR: Path-Based Multi-Hop Reasoning Incorporating Rule-Enhanced Reinforcement Learning and KG Embeddings3
A novel approach that combines KG embeddings and RL strategies for multi-hop reasoning called path-based multi-hop reasoning (PMHR), which provides not only improved reasoning accuracy and explainability but also optimized computational efficiency, thereby offering a robust solution for multi-hop reasoning.
- Node-dependent Semantic Search over Heterogeneous Graph Neural Networks3
To perform semantic search over HGNNs, this work proposes to develop semantic structures in terms of relation selection and connection selection, which could guide a task-relevant message flow, which justifies its effectiveness and efficiency.
- Which Matters Most in Making Fund Investment Decisions? A Multi-granularity Graph Disentangled Learning Framework3
A novel Multi-granularity Graph Disentangled Learning framework named MGDL is developed to effectively perform intelligent matching of fund investment products to attain stronger disentangled representations with specific semantics.
- A Comparison of Dynamic SPECT Coronary Flow Reserve with TIMI Frame Count in the Treatment of Non-Obstructive Epicardial Coronary Patients3
It is demonstrated that that CFR may be an alternative non-invasive method for identifying slow flow in non-obstructive coronary.
- 3
- KGNN: Distributed Framework for Graph Neural Knowledge Representation3
A novel framework called KGNN is developed to take full advantage of knowledge data for representation learning in the distributed learning system and jointly explore high-order structure and attribute information together in a fine-grained fashion and preserve the relation patterns in KGs, respectively.
- Gated Hypergraph Neural Network for Scene-Aware Recommendation3
This paper proposes a novel gated hypergraph neural network for Scene-aware Recommendation (SREC), which can fully leverage the scene prior knowledge and outperforms state-of-the-art methods on real industrial datasets.
- 3
- 3
- 3
- Schema-Rich Heterogeneous Network Mining3
This chapter introduces two data mining tasks on schema-rich heterogeneous network: link prediction and entity set expansion, and illustrates the challenges and potential solutions on mining this kind of more complex and popular heterogeneous networks.
- 3
- 3
- 3
- A Novel Algorithm for Hierarchical Community Structure Detection in Complex Networks3
A novel algorithm is introduced that searches for the hierarchical structure of networks and iteratively combines the similar communities with the elaborate design of community similarity and combination threshold to obtain reasonable hierarchical structure solutions.
- 3
- ORL-LO: Offline Reinforcement Learning for Pretraining and Finetuning in Logic Optimization2
This work proposes ORL-LO, a two-stage learning framework that leverages an offline reinforcement learning (RL) method to pretrain on expert trajectories and finetune on target test circuits, enabling efficient exploration of optimal synthesis flows.
- GraphLAMA: Enabling Efficient Adaptation of Graph Language Models with Limited Annotations2
This paper proposes GraphLAMA method, with its model backbone and learning schemes specialized for efficient tuning and inference, and achieves state-of-the-art (SOTA) performance with 4.91% absolute improvement in accuracy.
- The relationship between cognitive impairment and acoustic features in major depressive disorder: a longitudinal cohort study2
It is demonstrated that speech features, particularly spectral components, reflect cognitive impairment in MDD, offering promising non-invasive biomarkers for monitoring cognitive function in depression.
- 2
- 2
- 2
- A High-Efficient Method for Synthesizing Multiple Antenna Array Radiation Patterns Simultaneously Based on Convolutional Neural Network2
A high-efficient method that utilizes deep learning technology for synthesizing multiple antenna array radiation patterns simultaneously and offers the advantage of computational efficiency for simultaneous synthesis of multiple incoherent radiation patterns is proposed.
- 2
- 2
- 2
- 2
- Platforms and Practice of Heterogeneous Graph Representation Learning2
This chapter introduces OpenHGNN, a toolkit that can help to build HG models in a predesigned pipeline and presents the procedures with three well-known HG models that are HAN, the model that introduces attention mechanism to heterogeneous graph neural networks, RGCN, a model used to model multi-relational graphs with GCN, and HERec, a heterogeneous graph embedding method for recommendation.
- 2
- More Than One: A Cluster-Prototype Matching Framework for Zero-Shot Learning2
A novel cluster-prototype matching (CPM) strategy is introduced and a ZSL framework based on CPM is proposed that can significantly improve the performance of the basic ZSL models and help them achieve or beyond the state-of-the-art.
- HENA 20192
This workshop has an exciting program that spans a number of subareas, including: network construction and mining, network embedding, information diffusion, knowledge graph analysis, community detection, parallel computing for network analysis, and network analysis applications.
- 2
- Structure design for prediction system of radar electromagnetic environment on dynamic battlefield2
The prediction system of radar electromagnetic environment on dynamic battlefield helps people full understand the effect of external radar electromagnetic to combat action in different war situations and object-oriented technology is introduced in structure design for prediction system.
- 2
- 2
- 2
- Transferable Parasitic Estimation via Graph Contrastive Learning and Label Rebalancing in AMS Circuits1
CircuitGCL is a novel graph contrastive learning framework that integrates representation scattering and label rebalancing to enhance transferability across heterogeneous circuit graphs, and outperforms all state-of-the-art (SOTA) methods.
- 1
- 1
- AuCoGNN: Enhancing Graph Fairness Learning Under Distribution Shifts With Automated Graph Generation1
- 1
- 1
- 1
- 1
- An Effective Graph Learning based Approach for Temporal Link Prediction: The First Place of WSDM Cup 20221
This paper introduces the solution of AntGraph, which wins the 1st place in the WSDM Cup 2022, and analysis the theoretical upper-bound of the performance by removing temporal information, which implies that only structure and attribute information on the graph could achieve great performance.
- An Effective Sentiment Analysis Model for Tobacco Consumption1
This work presents a new large-scale multi-sentiment tobacco dataset by distilling effective consumer experience information from massive online reviews of tobacco consumption by proposing simple yet effective EHCRNN model, which combines the strengths of recent NLP advances.
- Attribute-Assisted Heterogeneous Graph Representation1
This chapter introduces three attribute-assisted heterogeneous graph representation models including heterogeneous graph attention network (HAN), heterogeneous graph propagation network (HPN), and heterogeneous graph structure learning (HGSL), which simultaneously utilize both complex structural information and rich attribute information to learn node representation.
- The 4th Workshop on Heterogeneous Information Network Analysis and Applications (HENA 2021)1
The 4th Workshop on Heterogeneous Information Network Analysis and Applications (HENA 2021) is co-located with the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining.
- 1
- 1
- 1
- Incorporating Instance Correlations in Distantly Supervised Relation Extraction1
This paper proposes a graph convolution network (GCN) model with an attention mechanism to improve relation extraction and demonstrates that the model significantly outperforms the compared baselines.
- 1
- Path-Based Ranking and Clustering1
The HRank method is introduced to evaluate the importance of multiple types of objects and meta paths, and the HeProjI algorithm is presented to solve the heterogeneous network projection and integration of clustering and ranking tasks.
- Prototype System Based on Heterogeneous Network1
This chapter introduces two prototype systems for recommendation and gives a brief review on other systems based on heterogeneous networks.
- Fusion Learning on Heterogeneous Social Networks1
The heterogeneous information generated by users’ social activities in the multiple aligned social networks provides social network practitioners and researchers with the opportunities to study individual user’s social behaviors across multiple social platforms simultaneously.
- 1
- 1
- VisNetMiner: An Integration Tool for Visualization and Analysis of Networks1
VisNetMiner is an integration tool for visualization and analysis of networks, which aims at integrating data preprocessing, modeling, analysis and visualization, which can analyze different kinds of data and builds a clear architecture for networkAnalysis and visualization.
- A Multi-objective optimal evolutionary algorithm based on tree-ranking1
Experiments prove that the proposed new method using tree structure to express the relationship of solutions can reach the Pare-to front, retain the diversity of the population, and use less time.
- –
- –
- –
- Riemannian Graph Tokenizer for Structural Knowledge Transfer–
A Riemannian Graph Tokenizer (RGT) is proposed that bridges the structural knowledge and quantized representations to support cross-domain structural knowledge transfer and significantly enhances structural knowledge transferability across graph domains.
- Hyper-KGGen: A Skill-Driven Knowledge Extractor for High-Quality Knowledge Hypergraph Generation–
Hyper-KGGen is a skill-driven framework that reformulates extraction as a dynamic skill-evolving process that significantly outperforms strong baselines, validating that evolved skills provide substantially richer guidance than static few-shot examples in multi-scenario settings.
- –
- –
- Masked Language Models are Good Heterogeneous Graph Generalizers–
A simple yet effective Masked Language Modeling-based method that introduces metapath-based textual sequences instead of HG tokens to extract structural and semantic information inherent in HGs, and designs customized textual templates to unify different graph tasks into a coherent cloze-style'mask'token prediction paradigm.
- –
- –
- Full-Atom Protein-Protein Interaction Prediction via Atomic Equivariant Attention Network–
This paper proposes a novel model, MEANT, designed to adaptively extract atom-level geometric information from varying numbers of atoms within different residues for PPI prediction, and significantly outperforms state-of-the-art methods on three typical PPI prediction tasks.
- –
- Duplicate Multi-modal Entities Detection with Graph Contrastive Self-training Network–
This work proposes a novel graph contrastive self-training network model, named CT-GNN, for duplicate multi-modal entities detection with extreme label sparsity, and designs a layer contrastive module to effectively exploit implicit label information, as well as a pseudo labels extension module to determine label boundary.
- –
- –
- –
- –
- –
- –
- –
- –
- Heterogeneous Graph Representation for Text Mining–
This chapter focuses on summarizing the heterogeneous graph representation applications on text mining, and introduces several heterogeneous graph-based text mining methods, including HGAT for short text classification, GUND and GNewsRec for news recommendation.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Structure-Preserved Heterogeneous Graph Representation–
This chapter introduces several works focusing on structure preservation, which successfully depict the rich semantics and complex heterogeneity, and effectively support downstream tasks ineterogeneous graph.
- –
- EasyGML–
EasyGML is presented, a fully-functional and easy-to-use platform for general AI practitioners to apply out-of-the-box GML models in industrial scenarios, and provides a user-friendly web-based GUI for users to build their GML workflows simply by connecting several components together, without any coding.
- NEAR: Normalized Network Embedding with Autoencoder for Top-K Item Recommendation–
This work proposes a personalized top-k item recommendation method that jointly considers accuracy and diversity, called Normalized Network Embedding with Autoencoder for Personalized Top-K Item Recommendation, namely NEAR, and compares it with some state-of-the-art algorithms on the DBLP and MovieLens1M datasets.
- A Rapid Localization Method of Radiation Sources Used for Multi-Sensor Networks–
A rapid localization method of radiation sources used for multi-sensor networks is provided, which can accomplish the target location of the radiation source effectively and improve the localization accuracy.
- Evolutionary multiobjective optimization for multilabel learning–
In this chapter, two works in evolutionary multiobjective optimization for multilabel learning are presented: in the first work, the author directly optimize multi objective functions based the multiobjectives optimization framework; in the second work, a set of accurate but diverseMultilabel learners are generated with the ensemble learning framework.
- –
- –
- –
- –
- –
- High-Dimensional Data Visualization Based on User Knowledge–
A high-dimensional data visualization method based on user knowledge that can derive dimensions aligned with user knowledge to reorganize data, then it uses scatter-pie plot matrix, an extension of scatterplot matrix, to visualize the reorganized data.
- A Novel Index Structure for Multi-key Search–
This work designs an index structure called D-Tree that supports sequential multi-key scan which is slow and energy-consuming, and designed to address the problem.
- –
- Recursion-removal Method Among Multi-function–
Experiments prove that this method can successfully remove the recursion of dominating tree and double the time efficiency before recursion-removal.
- A Posteriori Approach for Community Detection–
This paper proposes a posteriori decision approach for community detection in complex network that is able to obtain many candidate solutions, which effectively avoids the resolution limit existing in priori decision approaches.
- Fast fault location algorithm based on node betweenness–
The heuristic fault node location algorithm based on node's betweenness is proposed in this paper to promote fault location performance of traditional SNMP Polling in the time of finding fault nodes in real networks of star topology and mixed topology.
- –
- –
- Designation and Realization of Softswitch-based Remote Video-monitoring System–
The architecture of softswitch-based Next Generation Network (NGN) system and Session Initiation Protocol (SIP) are studied briefly, and the problems on the openness and extensibility of normal remote video-monitoring system (RVMS) are analyzed.
- Research on Multi-agent Reinforcement Learning with Sharing Experience–
Experimental results show that sharing state space can expedite the process of multi-agent reinforcement learning and the smaller the state space is, the faster Q learning algorithm convergence will be.
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-15. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.