基于大語(yǔ)言模型的知識(shí)圖譜構(gòu)建及在災(zāi)害應(yīng)急中的應(yīng)用

打開(kāi)文本圖片集

中圖分類號(hào)：TP391 文獻(xiàn)標(biāo)識(shí)碼：A 文章編號(hào)：2096-4706（2025）08-0146-07

Abstract：China isoneofthecountries withthe mostserious naturaldisasters inthe world.Typhoons，arthquakesfoods and ther disasters ocur frequently，posing a huge threat tothesafetyof people'slives and property.Facing thecomplexityand urgency ofdisasteremergencymanagement，Knowledge Graphtechnology has graduallybecomearesearch hotspot inthis field byvirtueofitsadvantagesiniformationintegration，elationsipingandknowledgereasonig.However，theeterogeity andqualityofisasterdataarenotufor，hichliitstheabilityoftraditioalNaturalLanguageProcessngmethodstotract Knowledge Graph information.Therefore，this paper proposes aLarge Language Model knowledge extraction method based on distributedpromptstrategy.Troughrequirementhints，domainknowledgehintsandfew-shothints，theLargeLanguageModel can automatically extract entitiesand relationships from unstructured data， which improves the automation and accuracy of KnowledgeGraphconstruction.InordertoverifytheefectofKnowledgeGraphbasedonLargeLanguageModelinpractical application，thispapertakes thetyphoondisasteremergencyplanasanexampletocostruct thecoresponding Knowledge Graph and its intelligent question and answer system.

Keywords：Knowledge Graph;Large Language Model; knowledge extraction; inteligent questionandanswer

0 引言

災(zāi)減災(zāi)要求。（剩余8910字）