HELEA: Hard-Negative Benchmark and LLM-based Reranking for Robust Entity Alignment
For researchers in knowledge graph fusion, this work provides a more challenging benchmark and a robust method to handle hard negatives, though the approach is incremental.
Entity alignment benchmarks often let models exploit name overlap, so the authors created hard-negative benchmarks (DW-HN29K, DY-HN27K) with same-name entities that refer to different objects, and proposed HELEA, a two-stage framework with encoder retrieval and LLM reranking. HELEA achieves F1 0.967 on DW-HN29K while maintaining Hit@1 0.993 on standard DW-15K.
Entity Alignment (EA) is essential for knowledge graph (KG) fusion, but existing benchmarks often allow models to exploit name overlap rather than relational structure. This makes it difficult to evaluate whether models can reject same-name entities that refer to different real-world objects. Our primary contribution is a same-name hard-negative augmentation strategy that simultaneously yields quality-controlled evaluation benchmarks (DW-HN29K, DY-HN27K) and augmented training corpora (DW-Train, DY-Train), by mining same-name but distinct entity pairs from KG name-collision groups. We further introduce HELEA, a two-stage framework integrating (i) entity encoder retrieval trained on hard-negative-augmented training corpora with 1-hop KG context, and (ii) LLM-based reranking without additional training. Experiments show that name-dependent baselines collapse to near-random performance on our hard-negative benchmarks, while HELEA achieves F1 0.967 on DW-HN29K while maintaining Hit@1 0.993 on standard DW-15K.