Research on log text fault association rule mining based on GPT network

LI Bing; YE Qingwei; WANG Yuxi; HE Zhenqin; WANG Xiaodong

doi:10.11959/j.issn.1000-0801.2025207

您当前的位置：

首页 >

文章列表页 >

Research on log text fault association rule mining based on GPT network

Research and Development | 更新时间：2026-01-08

- Research on log text fault association rule mining based on GPT network
- Telecommunications Science Vol. 41, Issue 12, Pages: 100-115(2025)
- 作者机构：
  
  宁波大学信息科学与工程学院，浙江宁波 315211
- 作者简介：
- 基金信息：
  
  Industry-University-Research Cooperation Project of Zhejiang Province(062400020);Large-Scale Horizontal Project Network Operation and Maintenance Platform Research and Development Project(HK2022000189)
- DOI：10.11959/j.issn.1000-0801.2025207
  CLC： TN915
- Received：24 March 2025，
  
  Revised：2025-05-12，
  
  Accepted：04 June 2025，
  
  Published：20 December 2025
- 稿件说明：
移动端阅览
李冰,叶庆卫,王雨晞等.基于GPT网络的日志文本故障关联规则挖掘研究[J].电信科学,2025,41(12):100-115.

LI Bing,YE Qingwei,WANG Yuxi,et al.Research on log text fault association rule mining based on GPT network[J].Telecommunications Science,2025,41(12):100-115.
李冰,叶庆卫,王雨晞等.基于GPT网络的日志文本故障关联规则挖掘研究[J].电信科学,2025,41(12):100-115. DOI： 10.11959/j.issn.1000-0801.2025207.

LI Bing,YE Qingwei,WANG Yuxi,et al.Research on log text fault association rule mining based on GPT network[J].Telecommunications Science,2025,41(12):100-115. DOI： 10.11959/j.issn.1000-0801.2025207.

摘要

日志文本异常检测及故障关联规则挖掘是现代系统运维和故障诊断中的关键环节，对于提升系统可靠性和降低运维成本具有重要意义。针对传统人工监控方式难以应对系统复杂性和规模增长的问题，提出了一种基于深度学习的通信网日志信息关联规则挖掘方法。该方法首先从通信网日志信息中提取主机IP地址、时间和故障类型等关键信息，并采用多种经典关联规则算法进行融合，生成关联规则标签，构建深度学习网络的数据集及标签集。然后，引入逐层学习机制，设计了一种高效的深度学习模型，在构建的数据集上进行训练，获得专门用于挖掘通信网故障事件之间关联关系的深度学习网络模型。在GAIA大型日志数据集上的实验结果表明，新算法展现出显著优势：在准确率方面达到64.2%，较FP-Growth算法（22.1%）提升190.5%，较Eclat算法（43.7%）提升46.9%，较Quant Matrix Miner算法（37.6%）提升70.7%；在运行时间方面仅需要11 s，较FP-Growth算法提速138.2%，较图神经网络（graph neural network，GNN）提速195.0%，同时保持了与Apriori算法相近的处理速度。此外，新算法在支持度（0.42）和置信度（0.75）等关键指标上也表现出良好的均衡性。该方法为故障事件关联规则的智能挖掘提供了新的工具，具有一定的实用价值和应用前景。

Abstract

Log anomaly detection and fault association rule mining are critical components in modern system operation and maintenance

as well as fault diagnosis

playing a significant role in enhancing system reliability and reducing operational costs. To address the challenges posed by the increasing complexity and scale of systems

which traditional manual monitoring methods struggle to handle

a deep learning-based method for mining association rules from communication network logs was proposed. The method first extracted key information such as host IP addresses

timestamps

and fault types from communication network logs. Subsequently

it integrated multiple classic association rule algorithms to generate association rule labels

constructing a dataset and label set for the deep learning network. Subsequently

a hierarchical learning mechanism was introduced

and an efficient deep learning model was designed to train on the constructed dataset

resulting in a specialized deep learning network model for mining association relationships among communication network fault events. The experimental outcomes derived from the comprehensive GAIA log dataset underscore the pronounced superiority of the proposed method. It achieved an accuracy of 64.2%

representing a considerable enhancement of 190.5% over the FP-Growth algorithm’s performance of 22.1%

46.9% over the Eclat algorithm’s 43.7%

and 70.7% over Quant Matrix Miner’s 37.6%. With respect to operational time

the method exhibited remarkable speed

requiring only 11 seconds for execution

thereby outpacing the FP-Growth algorithm by 138.2% and the graph neural network (GNN) by 195.0%

while maintaining a comparable processing velocity to the Apriori algorithm. Furthermore

the method demonstrated a well-balanced performance in key metrics including support (0.42) and confidence (0.75). This method provides a new tool for the intelligent mining of fault event association rules

offering practical value and promising application prospects.

关键词

Keywords

references

樊志强 , 刘日昇 . 网络设备智能化管理的研究与应用 [J ] . 物联网技术 , 2023 , 13 ( 8 ): 51 - 55 .

FAN Z Q , LIU R S . Research and application of intelligent management of network equipment [J ] . Internet of Things Technologies , 2023 , 13 ( 8 ): 51 - 55 .

GIRACCA D M , PIRES F L , BARÁN B , et al . A deep learning approach for anomaly detection for industrial control systems [C ] // Proceedings of the 2024 L Latin American Computer Conference (CLEI) . Piscataway : IEEE Press , 2024 : 1 - 10 .

YANG J , SHENG Y Q , WANG J L , et al . CAGCN: centrality-aware graph convolution network for anomaly detection in industrial control systems [J ] . Journal of Computer Science and Technology , 2024 , 39 ( 4 ): 967 - 983 .

CHOE H O , LEE M H . Artificial intelligence-based fault diagnosis and prediction for smart farm information and communication technology equipment [J ] . Agriculture , 2023 , 13 ( 11 ): 2124 .

SHETH V , TRIPATHI U , SHARMA A . A comparative analysis of machine learning algorithms for classification purpose [J ] . Procedia Computer Science , 2022 , 215 : 422 - 431 .

WANG K L . Attention-based BiGRU-CNN for system log anomaly detection [C ] // Proceedings of the 2024 4th International Symposium on Computer Technology and Information Science (ISCTIS) . Piscataway : IEEE Press , 2024 : 151 - 156 .

DONG Y , ZHANG Z W , PENG H Y , et al . New perspective on progressive GANs distillation for one-class anomaly detection [J ] . Journal of Imaging Science and Technology , 2023 , 67 ( 6 ): 1 - 11 .

LI Z , SHI J Y , VAN LEEUWEN M . Graph neural networks based log anomaly detection and explanation [C ] // Proceedings of the 2024 IEEE/ACM 46th International Conference on Software Engineering: Companion Proceedings . New York : ACM , 2024 : 306 - 307 .

LOU Z Y , CHAI X L , SHANG C , et al . Generalizable and robust log anomaly detection based on Transformer [C ] // Wireless Sensor Networks . Singapore : Springer , 2025 : 168 - 177 .

WANG C M , FU A Q , LI W D , et al . Intelligent identification of hidden dangers in hydrogen pipeline transmission station using GWO-optimized Apriori algorithm [J ] . Energies , 2024 , 17 ( 18 ): 4539 .

XIAO Y , MENG L H , ZHANG Y R , et al . An improved FP-Growth algorithm with time decay factor and element attention weight [C ] // Proceedings of the 2024 IEEE 4th International Conference on Power, Electronics and Computer Applications (ICPECA) . Piscataway : IEEE Press , 2024 : 860 - 865 .

PAPPA L , KARVELIS P , STYLIOS C . Exploring the diverse world of SAX-based methodologies [J ] . Data Mining and Knowledge Discovery , 2024 , 39 ( 1 ): 4 .

WANG S Z , JIANG R X , WANG Z Q , et al . Deep learning-based anomaly detection and log analysis for computer networks [J ] . arXiv preprint , 2024 : 2407 .05639.

JIN G , JING C , ZONG T Q , et al . Data redundancy elimination and noise processing via large language model prompt engineering [C ] // Proceedings of the 2024 10th International Conference on Big Data and Information Analytics (BigDIA) . Piscataway : IEEE Press , 2024 : 818 - 825 .

MARJAI P , KISS A . The usage of template mining in log file classification [J ] . IEEE Access , 2024 , 12 : 96378 - 96386 .

DU X Y , XU C , LI L , et al . Multigranularity feature automatic marking-based deep learning for anomaly detection of industrial control systems [J ] . IEEE Open Journal of Instrumentation and Measurement , 2024 , 3 : 2500110 .

CHAI X L , ZHANG H , ZHANG J , et al . Log sequence anomaly detection based on template and parameter parsing via BERT [J ] . IEEE Transactions on Dependable and Secure Computing , 2025 , 22 ( 2 ): 1150 - 1167 .

LIU M , HU S , ZHANG J , et al . Methods for identifying complex lithologies from log data based on machine learning [J ] . Unconventional Resources , 2023 , 3 : 20 - 29 .

LI Z L , TU X Z , GAO H , et al . LogCSS: log anomaly detection based on BERT-CNN with context-semantics-statistics features [J ] . Journal of Intelligent & Fuzzy Systems , 2024 , 46 ( 4 ): 7659 - 7676 .

王雨晞 , 叶庆卫 , 周鹏 , 等 . 日志信息驱动的计算机网络节点故障预测研究 [J ] . 电信科学 , 2024 , 40 ( 8 ): 11 - 22 .

WANG Y X , YE Q W , ZHOU P , et al . Research on fault prediction of computer network nodes driven by log information [J ] . Telecommunications Science , 2024 , 40 ( 8 ): 11 - 22 .

GUO H X , YUAN S H , WU X T . LogBERT: log anomaly detection via BERT [C ] // Proceedings of the 2021 International Joint Conference on Neural Networks (IJCNN) . Piscataway : IEEE Press , 2021 : 1 - 8 .

BAO F G , MAO L H , ZHU Y L , et al . An improved evaluation methodology for mining association rules [J ] . Axioms , 2022 , 11 ( 1 ): 17 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Research on fault prediction of computer network nodes driven by log information

Policy conflict detection in software defined network by using deep learning

Research on SRv6 Network Traffic Anomaly Detection Based on ResNet18

Research on anomaly detection method of power grid control system based on CVE-ReS-GRU algorithm

Related Author

YE Qingwei

ZHOU Peng

Chuanhuang LI

Cheng CHENG

Xiaoyong YUAN

Lijie CEN

Weiming WANG

Zhao Ruifeng

Related Institution

LiLAB,University of Florida,Gainesville

School of Information and Electrical Engineering,Zhejiang Gongshang University

NARI-TECH Nanjing Control Systems Co., Ltd.

Electric Power Dispatching Control Center of Guangdong Power Grid Co., Ltd.

Guoneng Zhejiang Beilun First Power Generation Co., Ltd.

⁰