面向電信領域的大模型提示詞工程測評

打開文本圖片集

中圖分類號：TP182 文獻標識碼：A 文章編號：2096-4706（2025）12-0123-06

Prompt Engineering Evaluation of Large Language Model for the TelecommunicationsDomain

FAN Wenbin1， WANG Yanyan1， WANG Yingying1， XU Yin1， SONG Qi2 （1.KnowledgeComputing InteligenceLaboratory，GuoChuang CloudTechnologyCo.，Ltd.，Hefei3oo88，China; 2.SchoolofComputerScienceandTechnologyUniversityofScienceandTechnologyofChina，Hefei23o027，China）

Abstract： A prompt evaluation system of Large Language Model （LLM） forthe telecommunications domain is proposed toadresste isuesofincomplete evaluationofpromptparameters inpromptengineeringresearchand thelackofconsideration forthecomplexityinrealproductionsenariosofevaluationmethod.Tothisend，fivedatasets inthe telecommunications domainareconstructed，coveringthree majortasksofsntimenttextclasification，customersrvice intentrecogniionnd knowledge-basedquestionanswering.Subsequentlypromptparametersarecategorized intofourdimensionsofole，lngth， tone，andorder，andthe impactofthesediferentdimensionsontheperformanceofsixLLMsissystematicallyevaluated.The researchresults indicatethata well-esigned promptcansignificantlyimprovemodel performanceonthethreemajortasks inthe telecommunications domain.

Keywords： Large Language Model; prompt enginering; model performance optimization; telecommunication domain; Jatural Language Processing

0 引言

近年來，人工智能技術(shù)迅猛發(fā)展，其中大語言模型（LargeLanguageLodels，LLMs）作為自然語言處理領域的核心技術(shù)，受到了廣泛關(guān)注。（剩余8179字）