基于LLMs的中国人民志愿军人物档案知识自动抽取与智慧问答研究

打开文本图片集

Abstract：To address the demand forintelligent development and utilization of Chinese People'sVolunteersArmy personnel archives， the pa perexplores a method forautomated knowledge extraction and intelligent question answering based on LLMs，aiming to support the digital and intelligent inheritance of the spirit of the War to Resist US Aggression and Aid Korea. It proposesan LLM- based framework for automated knowledge extraction and intelligent QA of CPVs personnelarchives.Takingthearchivesfromthe Resist US Aggression and Aid Korea Memorial Museum as an example，and leveraging the OpenSPG-KAG professional domain knowledge service framework，we providesaone-stop solution for ontology declaration，automated knowledge extraction and mapping， graph construction，and intelligent QA.Through associative clue indexing，semantic knowledgeexpansion，and knowledge- augmented generation， the study discovers and analyzes multidimensional knowledgeaboutwarfigures，militaryactivities，andintelligent QA.Iteffectivelyuncovers complexrelationships such as comrade，fellow-townsman， superior-subordinate，and military unit.The paperestablishesa complete workflow from theo retical framework to practical validation，offering aninnovative path for the integrated and finegrained dynamic inheritanceofsimilararchives.

Keywords：Chinese People's Volunteers Army personnel archives; LLMs;Knowledge- augmented generation;Knowledge graph;Resist US Aggression and Aid Korea

习近平总书记多次强调红色档案利用的重要性，要求“把蕴含党的初心使命的红色档案保管好、利用好，把新时代党领导人民推进实现中华民族伟大复兴的奋斗历史记录好、留存好"。（剩余9101字）