Author:
Huang Kui,Nie Wen,Luo Nianxue
Abstract
Constructed emergency response scenarios provide a basis for decision makers to make management decisions, and the development of such scenarios considers earlier historical cases. Over the decades, the development of emergency response scenarios has mainly implemented the elements of historic cases to describe the grade and influence of an accident. This paper focuses on scenario construction and proposes a corresponding framework based on natural language processing (NLP) using text reports of marine oil spill accidents. For each accident, the original textual reports are first divided into sentence sets corresponding to the temporal evolution. Each sentence set is regarded as a textual description of a marine oil spill scenario. A method is proposed in this paper, based on parsing, named entity recognition (NER) and open information extraction (OpenIE) to process the relation triples that are extracted from the sentence sets. Finally, the relation triples are semantically clustered into different marine oil spill domains to construct scenarios. The research results are validated and indicate that the proposed scenario construction framework can be effectively used in practical applications.
Subject
Health, Toxicology and Mutagenesis,Public Health, Environmental and Occupational Health
Reference34 articles.
1. Oil Pollution Control;Zaide,2019
2. An investigation on absorption properties of exfoliated graphite for oil spill from Caspian Sea water;Ghasemi;Iran. Chem. Commun.,2019
3. A review of polymer nanofibres by electrospinning and their application in oil–water separation for cleaning up marine oil spills
4. Oil Tanker Spill Statistics 2019https://www.itopf.org/knowledge-resources/data-statistics/statistics
5. Oil Spill Science and Technology;Fingas,2011
Cited by
4 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献