Benchmarking Neural Embeddings for Link Prediction in Knowledge Graphs Under Semantic and Structural Changes

Recently, link prediction algorithms based on neural embeddings have gained tremendous popularity in the Semantic Web community, and are extensively used for knowledge graph completion. While algorithmic advances strongly focused efficient ways of learning embeddings, fewer attention has been drawn to different their performance robustness can be evaluated. In this work we propose an open-source evaluation pipeline, which benchmarks accuracy situations where graphs may experience semantic structural changes. We define relation-centric connectivity measures that allow us connect capacity structure graph. Such pipeline is especially important simulate expected frequently updated.