Respecting Temporal-Causal Consistency: Entity-Event Knowledge Graphs for Retrieval-Augmented Generation

2025-06-06

AI Summary

Retrieval-augmented generation (RAG) struggles with narrative documents (e.g., novels) due to difficulties in modeling temporal evolution, causal dependencies, and character consistency. Method: This paper proposes E²RAG, a dual-graph framework that introduces the first bipartite knowledge graph structure—separately encoding entity and event subgraphs—and explicitly models evolutionary relationships via temporal encoding and dynamic bipartite mapping. Contribution/Results: We release ChronoQA, the first benchmark specifically designed to evaluate narrative RAG on temporal reasoning, causal inference, and character consistency. Experiments show that E²RAG significantly outperforms mainstream unstructured RAG and KG-RAG baselines on ChronoQA, achieving up to a 27.4% absolute accuracy gain on causal and character-consistency questions. E²RAG establishes a novel, interpretable, and evolution-aware structured RAG paradigm for narrative understanding.

Technology Category

• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Natural Language Processing: Question Answering
• Planning, Routing, and Scheduling: Temporal Planning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Retrieval-augmented generation (RAG) based on large language models often falters on narrative documents with inherent temporal structures. Standard unstructured RAG methods rely solely on embedding-similarity matching and lack any general mechanism to encode or exploit chronological information, while knowledge graph RAG (KG-RAG) frameworks collapse every mention of an entity into a single node, erasing the evolving context that drives many queries. To formalize this challenge and draw the community's attention, we construct ChronoQA, a robust and discriminative QA benchmark that measures temporal, causal, and character consistency understanding in narrative documents (e.g., novels) under the RAG setting. We then introduce Entity-Event RAG (E^2RAG), a dual-graph framework that keeps separate entity and event subgraphs linked by a bipartite mapping, thereby preserving the temporal and causal facets needed for fine-grained reasoning. Across ChronoQA, our approach outperforms state-of-the-art unstructured and KG-based RAG baselines, with notable gains on causal and character consistency queries. E^2RAG therefore offers a practical path to more context-aware retrieval for tasks that require precise answers grounded in chronological information.

Problem

• Handling temporal structures in narrative documents for RAG
• Preserving evolving entity context in knowledge graph RAG
• Ensuring temporal and causal consistency in QA benchmarks

Innovation

• Dual-graph framework for entity-event separation
• Bipartite mapping preserves temporal-causal facets
• ChronoQA benchmark for temporal consistency evaluation