HyGRAIL: Cost-Aware and Evidence-Grounded Scientific Hypothesis Discovery over Knowledge Graphs

Scientific knowledge graphs organize entities and relations from literature but often remain incomplete. Missing typed links represent plausible scientific hypotheses, though discovery is difficult due to sparsity. While GNNs are efficient but unreliable and LLMs are knowledgeable but costly, HyGRAIL offers a cost-aware, evidence-grounded approach. More broadly, graphs are essential computer science data structures for modeling complex relationships, such as social networks or navigation systems, supporting fundamental algorithms like BFS, DFS, and Dijkstra's algorithm.