摘要
针对传统话务质检存在规则覆盖率低、人工依赖度高、隐性需求识别困难等问题,基于大语言模型(LLM)的深度语义理解和跨领域推理能力,结合多模态数据融合与声纹识别等技术,提出了大语言模型智能质检解决方案。通过大语言模型智能质检新型系统的设计与应用,突破了传统关键词匹配局限,解决了传统NLP质检技术的局限,实现了端到端的话务全维度分析,提升了质检场景覆盖率和商机识别准确率,降低了情感分析误差,为电信等高交互行业提供了可解释、可落地的智能质检范式,推动客户服务从被动响应转向价值创造。
Aiming at the problems of low rule coverage,high manual dependence,and difficulty in identifying implicit requirements in tradi⁃tional call quality inspection,an intelligent quality inspection solution based on Large Language Model(LLM)is proposed.By leveraging LLM′s deep semantic understanding and cross domain reasoning capabilities,we have broken through the limitations of traditional keyword matching and proposed a new framework for intelligent quality inspection systems based on large language models,effectively solving the industry challenges of implicit expression parsing and real-time warning.This solution utilizes multimodal data fusion and voiceprint recognition technology to achieve end-to-end comprehensive analysis of traffic,which can improve the coverage of quality inspection scenarios,accuracy of business opportunity recognition,reduce sentiment analysis errors,and provide an interpretable and implementable intelligent quality inspection paradigm for high in⁃teraction industries such as telecommunications,promoting customer service from passive response to value creation.
作者
胡靖
胡爽
何畅
HU Jing;HU Shuang;HE Chang(China Telecom Corporation Limited Sichuan Branch,Chengdu 610000,China)
出处
《通信与信息技术》
2025年第S1期102-104,116,共4页
Communication & Information Technology
关键词
大语言模型
多模态数据
话务智能质检
全维度分析
Large Language Model(LLM)
Multimodal data
Intelligent quality inspection of call traffic
Full-dimensional analysis
