Link Prediction
Link prediction aims to forecast missing connections in networks by analyzing existing relationships and node attributes. Current research heavily involves graph neural networks (GNNs), but also explores alternative approaches like traditional machine learning models and diffusion probabilistic models, often enhanced with techniques such as contrastive learning and data augmentation to improve accuracy and address issues like heterophily and long-tailed distributions. This field is crucial for advancing knowledge graph completion, recommendation systems, and other applications requiring the inference of relationships between entities, with ongoing efforts focused on improving model interpretability and fairness.
Papers
From Link Prediction to Forecasting: Information Loss in Batch-based Temporal Graph Learning
Moritz Lampert, Christopher Blöcker, Ingo Scholtes
GENIE: Watermarking Graph Neural Networks for Link Prediction
Venkata Sai Pranav Bachina, Ankit Gangwal, Aaryan Ajay Sharma, Charu Sharma
LinkGPT: Teaching Large Language Models To Predict Missing Links
Zhongmou He, Jing Zhu, Shengyi Qian, Joyce Chai, Danai Koutra