Abstract
Neural IR has advanced through two distinct paths: entity-oriented approachesleveraging knowledge graphs and multi-vector models capturing fine-grainedsemantics. We introduce QDER, a neural re-ranking model that unifies theseapproaches by integrating knowledge graph semantics into a multi-vector model.QDER's key innovation lies in its modeling of query-document relationships:rather than computing similarity scores on aggregated embeddings, we maintainindividual token and entity representations throughout the ranking process,performing aggregation only at the final scoring stage - an approach we call"late aggregation." We first transform these fine-grained representationsthrough learned attention patterns, then apply carefully chosen mathematicaloperations for precise matches. Experiments across five standard benchmarksshow that QDER achieves significant performance gains, with improvements of 36%in nDCG@20 over the strongest baseline on TREC Robust 2004 and similarimprovements on other datasets. QDER particularly excels on difficult queries,achieving an nDCG@20 of 0.70 where traditional approaches fail completely(nDCG@20 = 0.0), setting a foundation for future work in entity-awareretrieval.