# Event-Causal RAG: A Retrieval-Augmented Generation Framework for Long Video Reasoning in Complex Scenarios Source: [Event-Causal RAG: A Retrieval-Augmented Generation Framework for Long Video Reasoning in Complex Scenarios]() Collection index: [llms.txt]() Data updated: 2026-05-09T05:00:12.752Z Authors: Peizheng Yan, Yu Zhao, Liang Xie, Juntong Qi, Mingming Wang, Erwei Yin Published: 2026-05-07T13:01:28.000Z ## Abstract Recent large vision-language models have achieved strong performance on short- and medium-length video understanding, yet they remain inadequate for ultra-long or even infinite video reasoning, where models must preserve coherent memory over extended durations and infer causal dependencies across temporally distant events. Existing end-to-end video understanding methods are fundamentally limited by the $O(n^2)$ complexity of self-attention, while recent retrieval-augmented generation (RAG) approaches still suffer from fragmented clip-level memory, weak modeling of temporal and causal structure, and high storage and online inference costs. We present Event-Causal RAG, a lightweight retrieval-augmented framework for infinite long-video reasoning. Instead of indexing fixed-length clips, our method segments streaming videos into semantically coherent events and represents each event as a structured State-Event-State (SES) graph, capturing the event together with its surrounding state transitions. These graphs are merged into a global Event Knowledge Graph and stored in a dual-store memory that supports both semantic matching and causal-topological retrieval. On top of this memory, we design a bidirectional retrieval strategy to efficiently identify the most relevant event causal chains and provide them, together with the associated video evidence, to a backbone video foundation model for answer generation. Experiments on long-video understanding benchmarks demonstrate that Event-Causal RAG consistently outperforms strong clip-based retrieval baselines and long-context video models, particularly on questions requiring multi-event integration and causal inference across long temporal gaps, while also achieving improved memory efficiency and robust streaming performance. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Video Question Answering on NExT-QA 5-180s (test)](): Video Question Answering; NExT-QA 5-180s (test); Accuracy: 80.54. - [Video Question Answering on EventBench 60-1800s (test)](): Video Question Answering; EventBench 60-1800s (test); Contextual Accuracy: 62.28. - [Action Reasoning on Video-MME Long (test)](): Action Reasoning; Video-MME Long (1800s-3600s) (test); Accuracy: 46.67. - [Action Recognition on Video-MME Long (1800s-3600s) (test)](): Action Recognition; Video-MME Long (1800s-3600s) (test); Accuracy: 47.5. ## Original sources and implementation [Original paper]() [PDF]()