Entity-Duet Neural Ranking: Understanding the Role of Knowledge Graph Semantics in Neural Information Retrieval

TL;DR

This paper introduces EDRM, a neural ranking model that integrates knowledge graph semantics into neural information retrieval, improving generalization and effectiveness in search tasks.

Contribution

The paper proposes the Entity-Duet Neural Ranking Model (EDRM) that combines knowledge graph semantics with neural ranking, a novel approach in neural IR.

Findings

Knowledge graph semantics enhance neural ranking performance.

EDRM outperforms baseline models on commercial search logs.

Semantic integration improves model generalization.

Abstract

This paper presents the Entity-Duet Neural Ranking Model (EDRM), which introduces knowledge graphs to neural search systems. EDRM represents queries and documents by their words and entity annotations. The semantics from knowledge graphs are integrated in the distributed representations of their entities, while the ranking is conducted by interaction-based neural ranking networks. The two components are learned end-to-end, making EDRM a natural combination of entity-oriented search and neural information retrieval. Our experiments on a commercial search log demonstrate the effectiveness of EDRM. Our analyses reveal that knowledge graph semantics significantly improve the generalization ability of neural ranking models.