Dynamic Link Prediction with Temporally Enhanced Signed Graph Neural Networks

In three linesModular framework to enhance signed GNNs with temporal context. Introduces HCIM (Historical Context Integration Module) combining learnable temporal weighting, LSTM, and multi-head attention for link prediction in temporal signed networks. Evaluated on Bitcoin OTC, Bitcoin Alpha, Reddit with statistically significant improvements over static baseline.

Summary generated by Claude — human-verified