ma-compbio/Hyper-SAGNN

Hyper-SAGNN is a classic academic reference implementation for hypergraph neural networks, published around 2019. While it was innovative for its time by applying self-attention to hypergraph structures, the project currently suffers from high obsolescence risk. With 102 stars and zero recent velocity (age > 2000 days), it functions more as a historical benchmark than a living software project. The field of Graph ML has since consolidated into major frameworks like PyTorch Geometric (PyG) and Deep Graph Library (DGL), which now include more optimized and flexible hypergraph layers (e.g., HyperGCN, UniGNN, AllSet). The primary risk isn't from frontier labs like OpenAI, but from the natural evolution of academic research and the dominance of well-maintained graph libraries that have absorbed these techniques. For a technical investor, this represents a 'solved' niche where the specific codebase has little defensible value beyond its initial citation impact.