# TriHetGCN **Entity class:** Link-prediction model **Maturity:** Developing ## Definition **TriHetGCN** is a research model that incorporates triadic closure, heterogeneity, and balance or harmony signals into graph convolution for link prediction. Its appearance in the source dossier records an emerging method, not an established counterterrorism deployment. ## Relationships - **type of:** [[wiki/Graph Convolutional Network|Graph Convolutional Network]]. - **used for:** [[wiki/Link Prediction|Link Prediction]]. - **concepts:** triadic closure and heterogeneous graphs. ## Sources / Provenance - [arXiv — TriHetGCN, 2025](https://arxiv.org/html/2504.20492v1)