基于LangChain與DeepSeek的JavaScript教學(xué)輔助智能體的研究

打開文本圖片集

中圖分類號：TP311；TP391.4 文獻(xiàn)標(biāo)識碼：A 文章編號：2096-4706（2025）07-0076-07

Abstract： This paper proposes a localized inteligent teaching system based on LangChain and DeepSeek-7B-R1， aiming tosolve thechalengesofasynchronous programmingcomprehensionandDOMoperationerordetectionin JavaScript and jQuery teaching.Toaddresstheconflictsamong semanticunderstanding，privacysecurityandhardwarerequirements in traditional solutions，this paper designs a dual-engine architecture （rule pre-screening + LLMfine-tuning） thatintegratesdynamic ASTparsing andan enhancedRAGretrieval mechanism，constructinga teaching Knowledge Graphcovering 57typesof typical errors.Experiments demonstrate that thesystemachieves anaverage response time of 2.8 seconds on NVIDIA RTX 3060 device，and the teaching suggestion accuracy reaching 91 . 2 % ，witha 4 3 . 7 % improvement over conventionalLSTM approaches. Deployment in corporate internal training scenarios verifes its capabilityto enhance learner problem-solving eficiency by 41 % whilereducingteachermanual intervention by 76 % .Thisresearch provides the firsthigh-precisionand localized intelligent teaching framework supporting consumer-grade hardware for the programming education field， with open-source core components to lower development barriers.

Keywords： localized inteligent teachingsystem;LangChain rule engine; DeepSeek-7B-R1 fine-tuning; RAG mechanism; JavaScript Teaching Knowledge Graph;ASTdynamic parsing; consumer-grade hardware deployment; programming education efficiencyoptimization

0 引言

JavaScript作為Web開發(fā)的核心語言，占據(jù)GitHub項目總量的 2 8 . 3 % ，但其事件驅(qū)動、異步非阻塞等特性導(dǎo)致教學(xué)場景中普遍存在“概念理解-實踐應(yīng)用”斷層現(xiàn)象。（剩余10126字）