A5040000956- Advanced Graph Neural Networks
- Recommender Systems and Techniques
- Misinformation and Its Impacts
- Topic Modeling
- Crystallography and molecular interactions
- Ferroelectric and Negative Capacitance Devices
- Machine Learning in Materials Science
- Data Mining Algorithms and Applications
- Human Mobility and Location-Based Analysis
- Caching and Content Delivery
- Expert finding and Q&A systems
- Semantic Web and Ontologies
- Spam and Phishing Detection
- Computational Drug Discovery Methods
- Traffic Prediction and Management Techniques
Tsinghua University
2023-2024
Modeling the dynamics into graph neural networks (GNNs) contributes to understanding of evolution in dynamic graphs, which helps optimize temporal-spatial representations for real-world network problems. Empirically, GNN embedding requires additional temporal encoders, inevitably introduces learning parameters make GNNs oversized and inefficient. Furthermore, previous models are under same fixed term, causes short-temporal optimum. To address these issues, we propose WinGNN framework model...
In recent years, graph neural networks (GNNs) have made great progress in recommendation. The core mechanism of GNNs-based recommender system is to iteratively aggregate neighboring information on the user-item interaction graph. However, existing GNNs treat users and items equally cannot distinguish diverse local patterns each node, which makes them suboptimal recommendation scenario. To resolve this challenge, we present a node-wise adaptive network framework ApeGNN. ApeGNN develops...
Self-supervised learning (SSL) has recently achieved great success in mining the user-item interactions for collaborative filtering. As a major paradigm, contrastive (CL) based SSL helps address data sparsity Web platforms by contrasting embeddings between raw and augmented data. However, existing CL-based methods mostly focus on batch-wise way, failing to exploit potential regularity feature dimension. This leads redundant solutions during representation of users items. In this work, we...
Recently, molecular data mining has attracted a lot of attention owing to its great application potential in material and drug discovery. However, this task faces challenge posed by the scarcity labeled graphs. To overcome challenge, we introduce novel augmentation semi-supervised confidence-aware consistency regularization training framework for property prediction. The core our is strategy on graphs, named DropConn (Dropout Connection). generates pseudo graphs softening hard connections...
With the rapid proliferation of scientific literature, versatile academic knowledge services increasingly rely on comprehensive graph mining. Despite availability public graphs, benchmarks, and datasets, these resources often fall short in multi-aspect fine-grained annotations, are constrained to specific task types domains, or lack underlying real graphs. In this paper, we present OAG-Bench, a comprehensive, multi-aspect, human-curated benchmark based Open Academic Graph (OAG). OAG-Bench...
Graph Neural Networks (GNNs) are commonly used and have shown promising performance in recommendation systems. A major branch, Heterogeneous GNNs, models heterogeneous information by leveraging side for academic paper recommendations. These networks use message passing high-order propagation to learn representations users items. However, existing methods perform propagation, leading suboptimal representation learning. To address this issue, proposes a framework called MCAP, which uses...
Social bot detection is becoming a task of wide concern in social security. All along, the development technology hindered by lack high-quality annotated data. Besides, rapid AI Generated Content (AIGC) dramatically improving creative ability bots. For example, recently released ChatGPT [2] can fool state-of-the-art AI-text-detection method with probability 74%, bringing large challenge to content-based methods. To address above drawbacks, we propose Contrastive Learning-driven Bot Detection...