{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,15]],"date-time":"2026-05-15T09:12:37Z","timestamp":1778836357562,"version":"3.51.4"},"reference-count":40,"publisher":"MDPI AG","issue":"5","license":[{"start":{"date-parts":[[2026,5,2]],"date-time":"2026-05-02T00:00:00Z","timestamp":1777680000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Guangdong Emergency Management Science and Technology Program","award":["2025YJKJ001"],"award-info":[{"award-number":["2025YJKJ001"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Algorithms"],"abstract":"<jats:p>Accident reports provide a detailed account of environmental causes, unsafe human behaviors, and subsequent chain reactions. These records serve as essential resources for analyzing accident mechanisms and exploring potential risk patterns within production safety processes. Currently, Graph based Retrieval-Augmented Generation (RAG), which integrates Large Language Models (LLMs) with Knowledge Graphs (KGs), has emerged as a leading approach for complex causal question answering over extensive unstructured accident documentation. However, the application of this technology in the production safety domain still encounters two primary challenges. First, knowledge graph construction using a single granularity fails to capture fine-grained case details and macro-level standard systems. Second, traditional one-step retrieval paradigms lack the capacity to track deep causal chains or interpret the complex logic of multi-factor coupling. To address these limitations, we propose CausalAgent, a hierarchical graph-enhanced multi-agent framework for causal question answering in production safety accident reports. This framework innovatively combines a Hierarchical Causal Graph (HC-Graph) and a Multi-Agent Collaborative Reasoning (MACR) mechanism. Specifically, the HC-Graph employs a two-layer architecture that links a fine-grained instance layer with a national standard causation layer to resolve conflicts in semantic granularity. The MACR mechanism converts complex natural language queries into executable structured queries and logic verification steps through the sequential cooperation of four specialized agents, namely the Graph Parsing Agent, the Problem Analysis Agent, the Query Generation Agent, and the Reasoning Insight Agent. CausalAgent enables in-depth mining of accident causation mechanisms and provides scientific, robust and interpretable intelligent support for data-driven risk assessment and emergency decision-making. Experiments on real-world accident datasets demonstrate that CausalAgent achieves a 100.0% query execution rate and an 87.3% reasoning accuracy, outperforming the SOTA baseline by 45.2% in terms of absolute accuracy.<\/jats:p>","DOI":"10.3390\/a19050355","type":"journal-article","created":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T01:12:09Z","timestamp":1777857129000},"page":"355","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["CausalAgent: A Hierarchical Graph-Enhanced Multi-Agent Framework for Causal Question Answering in Production Safety Accident Reports"],"prefix":"10.3390","volume":"19","author":[{"given":"Tianyi","family":"Wang","sequence":"first","affiliation":[{"name":"School of Electronic and Information Engineering, South China University of Technology, Guangzhou 510640, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Tao","family":"Shen","sequence":"additional","affiliation":[{"name":"School of Electronic and Information Engineering, South China University of Technology, Guangzhou 510640, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zhiyuan","family":"Zhang","sequence":"additional","affiliation":[{"name":"School of Electronic and Information Engineering, South China University of Technology, Guangzhou 510640, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Shuangping","family":"Huang","sequence":"additional","affiliation":[{"name":"School of Electronic and Information Engineering, South China University of Technology, Guangzhou 510640, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1419-059X","authenticated-orcid":false,"given":"Huiguo","family":"He","sequence":"additional","affiliation":[{"name":"School of Electronic and Information Engineering, South China University of Technology, Guangzhou 510640, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Qingguang","family":"Chen","sequence":"additional","affiliation":[{"name":"Guangdong Academy of Safety Production and Emergency Management Science and Technology, Guangzhou 510060, China"},{"name":"Guangdong Provincial Science and Technology Collaborative Innovation Center for Production Safety, Guangzhou 510060, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Houqiang","family":"Yang","sequence":"additional","affiliation":[{"name":"Guangdong Academy of Safety Production and Emergency Management Science and Technology, Guangzhou 510060, China"},{"name":"Guangdong Provincial Science and Technology Collaborative Innovation Center for Production Safety, Guangzhou 510060, China"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1968","published-online":{"date-parts":[[2026,5,2]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","unstructured":"Dhinakaran, D., Priya, S., Pravin, A., and Ganesh, D. (2024). Precision Maintenance with PARM and Augmented Reality for Asset Optimization. Navigating the Augmented and Virtual Frontiers in Engineering, IGI Global.","DOI":"10.4018\/979-8-3693-5613-5.ch002"},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"213","DOI":"10.1002\/sdr.338","article-title":"Learning from incidents: From normal accidents to high reliability","volume":"22","author":"Cooke","year":"2006","journal-title":"Syst. Dyn. Rev."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"115048","DOI":"10.1016\/j.oceaneng.2023.115048","article-title":"Research in marine accidents: A bibliometric analysis, systematic review and future directions","volume":"284","author":"Cao","year":"2023","journal-title":"Ocean. Eng."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"75","DOI":"10.1016\/j.aap.2013.07.027","article-title":"Systems thinking, the Swiss Cheese Model and accident analysis: A comparative systemic analysis of the Grayrigg train derailment using the ATSB, AcciMap and STAMP models","volume":"68","author":"Underwood","year":"2014","journal-title":"Accid. Anal. Prev."},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"103265","DOI":"10.1016\/j.autcon.2020.103265","article-title":"Text mining-based construction site accident classification using hybrid supervised machine learning","volume":"118","author":"Cheng","year":"2020","journal-title":"Autom. Constr."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"103145","DOI":"10.1016\/j.autcon.2020.103145","article-title":"Automatically learning construction injury precursors from text","volume":"118","author":"Baker","year":"2020","journal-title":"Autom. Constr."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"101152","DOI":"10.1016\/j.aei.2020.101152","article-title":"Hazard analysis: A deep learning and text mining framework for accident prevention","volume":"46","author":"Zhong","year":"2020","journal-title":"Adv. Eng. Inform."},{"key":"ref_8","unstructured":"Comanici, G., Bieber, E., Schaekermann, M., Pasupat, I., Sachdeva, N., Dhillon, I., Blistein, M., Ram, O., Zhang, D., and Rosen, E. (2025). Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities. arXiv."},{"key":"ref_9","unstructured":"Grattafiori, A., Dubey, A., Jauhri, A., Pandey, A., Kadian, A., Al-Dahle, A., Letman, A., Mathur, A., Schelten, A., and Vaughan, A. (2024). The llama 3 herd of models. arXiv."},{"key":"ref_10","unstructured":"Yang, A., Li, A., Yang, B., Zhang, B., Hui, B., Zheng, B., Yu, B., Gao, C., Huang, C., and Lv, C. (2025). Qwen3 technical report. arXiv."},{"key":"ref_11","unstructured":"Liu, A., Feng, B., Xue, B., Wang, B., Wu, B., Lu, C., Zhao, C., Deng, C., Zhang, C., and Ruan, C. (2024). Deepseek-v3 technical report. arXiv."},{"key":"ref_12","unstructured":"Singh, A., Fry, A., Perelman, A., Tart, A., Ganesh, A., El-Kishky, A., McLaughlin, A., Low, A., Ostrow, A., and Ananthram, A. (2025). Openai gpt-5 system card. arXiv."},{"key":"ref_13","first-page":"9459","article-title":"Retrieval-augmented generation for knowledge-intensive nlp tasks","volume":"33","author":"Lewis","year":"2020","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref_14","unstructured":"Asai, A., Wu, Z., Wang, Y., Sil, A., and Hajishirzi, H. (2024, January 7\u201311). Self-RAG: Learning to retrieve, generate, and critique through self-reflection. Proceedings of the The Twelfth International Conference on Learning Representations (ICLR), Vienna, Austria."},{"key":"ref_15","doi-asserted-by":"crossref","unstructured":"Jiang, Z., Xu, F.F., Gao, L., Sun, Z., Liu, Q., Dwivedi-Yu, J., Yang, Y., Callan, J., and Neubig, G. (2023, January 6\u201310). Active retrieval augmented generation. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, Singapore.","DOI":"10.18653\/v1\/2023.emnlp-main.495"},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"3580","DOI":"10.1109\/TKDE.2024.3352100","article-title":"Unifying Large Language Models and Knowledge Graphs: A Roadmap","volume":"36","author":"Pan","year":"2024","journal-title":"IEEE Trans. Knowl. Data Eng."},{"key":"ref_17","unstructured":"Edge, D., Trinh, H., Cheng, N., Bradley, J., Chao, A., Mody, A., Truitt, S., Metropolitansky, D., Ness, R.O., and Larson, J. (2024). From local to global: A graph rag approach to query-focused summarization. arXiv."},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Guo, Z., Xia, L., Yu, Y., Ao, T., and Huang, C. (2024). Lightrag: Simple and fast retrieval-augmented generation. arXiv.","DOI":"10.18653\/v1\/2025.findings-emnlp.568"},{"key":"ref_19","first-page":"59532","article-title":"Hipporag: Neurobiologically inspired long-term memory for large language models","volume":"37","author":"Shu","year":"2024","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"108015","DOI":"10.1016\/j.ocecoaman.2025.108015","article-title":"Automated knowledge extraction from marine accident reports using large language models: Graph construction and evaluation","volume":"272","author":"Huang","year":"2026","journal-title":"Ocean. Coast. Manag."},{"key":"ref_21","doi-asserted-by":"crossref","first-page":"129947","DOI":"10.1016\/j.eswa.2025.129947","article-title":"GRAG-ProSafe QAS: Graph retrieval-augmented generation for process production safety intelligent question and answer system","volume":"298","author":"Zhang","year":"2026","journal-title":"Expert Syst. Appl."},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"151","DOI":"10.1007\/s42979-025-03695-x","article-title":"Intelligent IoT-Driven Advanced Predictive Maintenance System for Industrial Applications","volume":"6","author":"Dhinakaran","year":"2025","journal-title":"SN Comput. Sci."},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"11196","DOI":"10.1109\/TKDE.2022.3223858","article-title":"Complex knowledge base question answering: A survey","volume":"35","author":"Lan","year":"2022","journal-title":"IEEE Trans. Knowl. Data Eng."},{"key":"ref_24","unstructured":"Luo, L., Li, Y.F., Haffari, G., and Pan, S. (2024, January 7\u201311). Reasoning on Graphs: Faithful and Interpretable Large Language Model Reasoning. Proceedings of the The Twelfth International Conference on Learning Representations (ICLR), Vienna, Austria."},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Zhang, J., Zhang, X., Yu, J., Tang, J., Tang, J., Li, C., and Chen, H. (2022, January 22\u201327). Subgraph retrieval enhanced model for multi-hop knowledge base question answering. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1:\nLong Papers), Dublin, Ireland.","DOI":"10.18653\/v1\/2022.acl-long.396"},{"key":"ref_26","unstructured":"Sun, J., Xu, C., Tang, L., Wang, S., Lin, C., Gong, Y., Ni, L.M., Shum, H.Y., and Guo, J. (2023). Think-on-graph: Deep and responsible reasoning of large language model on knowledge graph. arXiv."},{"key":"ref_27","first-page":"132876","article-title":"G-retriever: Retrieval-augmented generation for textual graph understanding and question answering","volume":"37","author":"He","year":"2024","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref_28","doi-asserted-by":"crossref","unstructured":"Jiang, J., Zhou, K., Dong, Z., Ye, K., Zhao, W.X., and Wen, J.R. (2023, January 6\u201310). StructGPT: A General Framework for Large Language Model to Reason over Structured Data. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP), Singapore.","DOI":"10.18653\/v1\/2023.emnlp-main.574"},{"key":"ref_29","doi-asserted-by":"crossref","unstructured":"Luo, H., Haihong, E., Tang, Z., Peng, S., Guo, Y., Zhang, W., Ma, C., Dong, G., Song, M., and Lin, W. (2024, January 11\u201316). Chatkbqa: A generate-then-retrieve framework for knowledge base question answering with fine-tuned large language models. Proceedings of the Findings of the Association for Computational Linguistics: ACL 2024, Bangkok, Thailand.","DOI":"10.18653\/v1\/2024.findings-acl.122"},{"key":"ref_30","doi-asserted-by":"crossref","first-page":"80981","DOI":"10.1109\/ACCESS.2025.3567759","article-title":"Decoding the mystery: How can LLMs turn text into cypher in complex knowledge graphs?","volume":"13","author":"Mandilara","year":"2025","journal-title":"IEEE Access"},{"key":"ref_31","unstructured":"Smeros, P., Emonet, V., Wang, R., Sima, A.C., and de Farias, T.M. (2025). SPARQL-LLM: Real-Time SPARQL Query Generation from Natural Language Questions. arXiv."},{"key":"ref_32","unstructured":"(2022). State Administration for Market Regulation; Standardization Administration of China. Classification and Code of Hazardous and Harmful Factors in Production Process (Standard No. GB\/T 13861-2022)."},{"key":"ref_33","doi-asserted-by":"crossref","unstructured":"Sarkar, S., Vinay, S., Djeddi, C., and Maiti, J. (2021). Text Mining-Based Association Rule Mining for Incident Analysis: A Case Study of a Steel Plant in India. Pattern Recognition and Artificial Intelligence, Springer.","DOI":"10.1007\/978-3-030-71804-6_19"},{"key":"ref_34","doi-asserted-by":"crossref","first-page":"736","DOI":"10.1109\/TKDE.2025.3639599","article-title":"A Framework of Knowledge Graph-Enhanced Large Language Model Based on Global Planning","volume":"38","author":"Li","year":"2025","journal-title":"IEEE Trans. Knowl. Data Eng."},{"key":"ref_35","unstructured":"Hong, S., Zhuge, M., Chen, J., Zheng, X., Cheng, Y., Wang, J., Zhang, C., Wang, Z., Yau, S.K.S., and Lin, Z. (2023). MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework. arXiv."},{"key":"ref_36","unstructured":"Kuai, C., Li, Z., Rosen, B., Paal, S., Jafari, N., Briaud, J.L., Zhang, Y., Hashash, Y., and Zhou, Y. (2025). Knowledge-Grounded Agentic Large Language Models for Multi-Hazard Understanding from Reconnaissance Reports. arXiv."},{"key":"ref_37","unstructured":"(2017). Industrial Classification for National Economic Activities (Standard No. GB\/T 4754-2017). (In Chinese)."},{"key":"ref_38","unstructured":"(2025). State Administration for Market Regulation (SAMR); Standardization Administration of the People\u2019s Republic of China (SAC). Classification and Coding of Work Safety Accidents (Standard No. GB 6441-2025)."},{"key":"ref_39","unstructured":"Ministry of Emergency Management of the People\u2019s Republic of China (2026, March 08). Disaster and Accident Information, Available online: https:\/\/www.mem.gov.cn\/xw\/zhsgxx\/."},{"key":"ref_40","unstructured":"(2026, March 08). Kimi K2: Open Agentic Intelligence. Moonshot AI Technical Report. Available online: https:\/\/moonshotai.github.io\/Kimi-K2\/."}],"container-title":["Algorithms"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1999-4893\/19\/5\/355\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,15]],"date-time":"2026-05-15T08:56:11Z","timestamp":1778835371000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1999-4893\/19\/5\/355"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,2]]},"references-count":40,"journal-issue":{"issue":"5","published-online":{"date-parts":[[2026,5]]}},"alternative-id":["a19050355"],"URL":"https:\/\/doi.org\/10.3390\/a19050355","relation":{},"ISSN":["1999-4893"],"issn-type":[{"value":"1999-4893","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5,2]]}}}