REGENT: Relevance-Guided Attention for Entity-Aware Multi-Vector Neural Re-Ranking
This addresses the challenge of intelligent content selection in information retrieval for users dealing with lengthy, multi-faceted texts, establishing a new paradigm for entity-aware retrieval.
The paper tackled the problem of neural re-rankers struggling with complex information needs and long documents by introducing REGENT, a model that uses entities to guide attention, achieving up to 108% improvement over BM25 and outperforming strong baselines like ColBERT and RankVicuna.
Current neural re-rankers often struggle with complex information needs and long, content-rich documents. The fundamental issue is not computational--it is intelligent content selection: identifying what matters in lengthy, multi-faceted texts. While humans naturally anchor their understanding around key entities and concepts, neural models process text within rigid token windows, treating all interactions as equally important and missing critical semantic signals. We introduce REGENT, a neural re-ranking model that mimics human-like understanding by using entities as a "semantic skeleton" to guide attention. REGENT integrates relevance guidance directly into the attention mechanism, combining fine-grained lexical matching with high-level semantic reasoning. This relevance-guided attention enables the model to focus on conceptually important content while maintaining sensitivity to precise term matches. REGENT achieves new state-of-the-art performance in three challenging datasets, providing up to 108% improvement over BM25 and consistently outperforming strong baselines including ColBERT and RankVicuna. To our knowledge, this is the first work to successfully integrate entity semantics directly into neural attention, establishing a new paradigm for entity-aware information retrieval.