Research on the automatic generation method of telecom re-complaints report based on improved Transformer model

LIANG Weiming; XIAO Jun; MA Xiaoliang; XIN Sheng; XU Rongbin

doi:10.11959/j.issn.1000-0801.2025110

您当前的位置：

首页 >

文章列表页 >

Research on the automatic generation method of telecom re-complaints report based on improved Transformer model

Engineering and Application | 更新时间：2025-06-28

- Research on the automatic generation method of telecom re-complaints report based on improved Transformer model
- Telecommunications Science Vol. 41, Issue 6, Pages: 197-207(2025)
- 作者机构：
  
  1.中国电信股份有限公司广州分公司，广东广州 510620
  2.中数通信息有限公司，广东广州 510630
  3.马晓亮劳模与工匠人才创新工作室，广东广州 510620
- 作者简介：
- 基金信息：
- DOI：10.11959/j.issn.1000-0801.2025110
  CLC： TP399
- Received：11 December 2024，
  
  Revised：2025-04-07，
  
  Published：20 June 2025
- 稿件说明：
移动端阅览
梁伟明,肖军,马晓亮等.基于改进Transformer的电信重投报告自动生成方法研究[J].电信科学,2025,41(06):197-207.

LIANG Weiming,XIAO Jun,MA Xiaoliang,et al.Research on the automatic generation method of telecom re-complaints report based on improved Transformer model[J].Telecommunications Science,2025,41(06):197-207.
梁伟明,肖军,马晓亮等.基于改进Transformer的电信重投报告自动生成方法研究[J].电信科学,2025,41(06):197-207. DOI： 10.11959/j.issn.1000-0801.2025110.

LIANG Weiming,XIAO Jun,MA Xiaoliang,et al.Research on the automatic generation method of telecom re-complaints report based on improved Transformer model[J].Telecommunications Science,2025,41(06):197-207. DOI： 10.11959/j.issn.1000-0801.2025110.

摘要

在电信行业中，客户对未解决或处理不满意的投诉进行重复投诉的现象较为常见。手动生成重投报告不仅耗时且主观性较强，难以满足企业对高效性和一致性的要求。针对这一问题，提出了一种基于改进Transformer模型的自动化报告生成方法。该方法通过引入情绪嵌入，有效捕捉客户在对话中的情绪变化，改善了生成报告对客户态度和诉求的理解能力。同时，结合定制化位置编码，提升了模型对投诉时序信息的感知能力，从而增强了生成内容的时间逻辑性和细节完整性。实验结果表明，改进后的模型在BLEU（bilingual evaluation understudy）和ROUGE（recall-oriented understudy for gisting evaluation）指标上分别达到0.352和0.482，显著优于原始Transformer和其他对比模型。此外，与人工对比，工作效率提高了89%。生成的报告内容不仅更加准确贴合实际需求，还在语义细节与时序一致性上表现优异。

Abstract

In the telecommunications industry

repeated customer complaints about unresolved or unsatisfactory issues are a common challenge. Manually generating re-investment reports is not only time-consuming and prone to subjectivity but also fails to meet enterprise demands for efficiency and consistency. To address this issue

an automatic report generation method based on an improved Transformer model was proposed. This method introduced emotion embedding

enabling the model to effectively capture dynamic emotional changes in customer interactions and better understand customer attitudes and demands during the dialogue. Additionally

the incorporation of customized position encoding enhanced the model’s ability to perceive complaint time series information

significantly improving the time logic and detailed completeness of the generated content. Experimental results demonstrate that the proposed model achieves BLEU (bilingual evaluation understudy) and ROUGE (recall-oriented understudy for gisting evaluation) scores of 0.352 and 0.482

respectively

outperforming the original Transformer and other baseline models. Moreover

compared to manual efforts

the proposed model improves work efficiency by 89%. The generated reports not only align more accurately with real-world requirements but also exhibit superior performance in semantic detail and time sequence consistency.

关键词

Keywords

references

曹璐 . 基于价值链的移动运营服务质量管理研究 [D ] . 武汉 : 武汉理工大学 , 2012 .

CAO L . Research on service quality management of mobile operation based on value chain [D ] . Wuhan : Wuhan University of Technology , 2012 .

李季 , 张帅 , 周静 . 服务与需求的匹配度对客户流失的影响研究: 基于电信行业的客户数据实验 [J ] . 管理评论 , 2020 , 32 ( 5 ): 192 - 204 .

LI J , ZHANG S , ZHOU J . The impact of matching between service and demand on customer churn: evidence from telecommunication industry [J ] . Management Review , 2020 , 32 ( 5 ): 192 - 204 .

袁琳 , 孙巍 , 马晓敏 , 等 . 图模型框架下的报道性新闻自动摘要方法研究 [J ] . 图书情报工作 , 2024 , 68 ( 17 ): 122 - 135 .

YUAN L , SUN W , MA X M , et al . Research on automatic summary methods for reportable news under the graph model framework [J ] . Library and Information Service , 2024 , 68 ( 17 ): 122 - 135 .

SUTSKEVER I , VINYALS O , LE Q V . Sequence to sequence learning with neural networks [J ] . arXiv preprint , 2014 : 1409. 3215 v 3 .

VASWANI A , SHAZEER N , PARMAR N , et al . Attention is all you need [J ] . arXiv preprint , 2017 : 1706 .03762.

陈嘉鸿 , 黄国恒 , 谭喆 . 基于跨模态差异注意力的医学报告生成 [J ] . 广东工业大学学报 , 2025 , 42 ( 1 ): 70 - 78 .

CHEN J H , HUANG G H , TAN Z . Cross-modal discrepancy attention network for medical report generation [J ] . Journal of Guangdong University of Technology , 2025 , 42 ( 1 ): 70 - 78 .

谭金源 , 刁宇峰 , 杨亮 , 等 . 基于BERT-SUMOPN模型的抽取-生成式文本自动摘要 [J ] . 山东大学学报(理学版) , 2021 , 56 ( 7 ): 82 - 90 .

TAN J Y , DIAO Y F , YANG L , et al . Extractive-abstractive text automatic summary based on BERT-SUMOPN model [J ] . Journal of Shandong University (Natural Science) , 2021 , 56 ( 7 ): 82 - 90 .

PAREKHR , PATELN P , THAKKARN , et al . DL-GuesS: deep learning and sentiment analysis-based cryptocurrency price prediction [J ] . IEEE Access , 2022 , 10 : 35398 - 35409 .

GHOSH P , SAMANTA O , GOTO T , et al . Sales forecasting of overrated products: fine tuning of customer’s rating by integrating sentiment analysis [J ] . IEEE Access , 2024 , 12 : 69578 - 69592 .

BORGA , BOLDTM . Using VADER sentiment and SVM for predicting customer response sentiment [J ] . Expert Systems with Applications , 2020 , 162 : 113746 .

DEVLIN J , CHANG M W , LEE K , et al . BERT: pre-training of deep bidirectional transformers for language understanding [J ] . arXiv preprint , 2018 : 1810 .04805.

张涛源 , 谢新林 , 谢刚 , 等 . 融合Transformer的带钢缺陷实时检测算法 [J ] . 计算机工程与应用 , 2023 , 59 ( 16 ): 232 - 239 .

ZHANG T Y , XIE X L , XIE G , et al . Real-time strip steel defect detection algorithm fused with Transformer [J ] . Computer Engineering and Applications , 2023 , 59 ( 16 ): 232 - 239 .

LIU F C , GAO C Q , CHEN F , et al . Infrared small and dim target detection with transformer under complex backgrounds [J ] . IEEE Transactions on Image Processing , 2023 , 32 : 5921 - 5932 .

潘芳 , 张会兵 , 董俊超 , 等 . 基于高效Transformer的中文在线课程评论方面情感分析 [J ] . 计算机科学 , 2021 , 48 ( S1 ): 264 - 269 .

PAN F , ZHANG H B , DONG J C , et al . Aspect sentiment analysis of Chinese online course review based on efficient Transformer [J ] . Computer Science , 2021 , 48 ( S1 ): 264 - 269 .

BELTAGY I , PETERS M E , COHAN A . Longformer: the long-document transformer [EB ] . 2020 .

廖俊伟 . 深度学习大模型时代的自然语言生成技术研究 [D ] . 成都 : 电子科技大学 , 2023 .

LIAO J W . Research on natural language generation technology in the era of deep learning big model [D ] . Chengdu : University of Electronic Science and Technology of China , 2023 .

姜雨娇 , 黄铝文 , 荚子萌 . 基于IMGRU-Seq2Seq的自动问答方法研究 [J ] . 计算机应用与软件 , 2024 , 41 ( 6 ): 215 - 222, 256 .

JIANG Y J , HUANG L W , JIA Z M . Automatic question answering method based on IMGRU-Seq2Seq [J ] . Computer Applications and Software , 2024 , 41 ( 6 ): 215 - 222, 256 .

Views

504

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Research on network intrusion detection based on fusion Transformer and Inception

A modulation recognition model driven by lightweight knowledge distillation in the complex number domain

GMTBLC: a deep learning-based bi-modal network traffic classification method

A robust encrypted traffic identification scheme based on graph neural network

Related Author

Song Zhenfeng

Wang Jiaxing

Zhang Wanqi

SONG Zhenfeng

WANG Jiaxing

ZHANG Wanqi

ZHOU Jin

GAO Shuo

Related Institution

College of Information Network Security, People’s Public Security University of China

College of Information Network Security, People's Public Security University of China

School of Information Science and Technology, Tianjin University of Finance and Economics

School of Information Engineering, Dalian University

University of Science and Technology of China

⁰