Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MicLog: Towards Accurate and Efficient LLM-based Log Parsing via Progressive Meta In-Context Learning

Jianbo Yu, Yixuan Li, Hai Xu, Kang Xu, Junjielong Xu, Zhijing Li, Pinjia He, Wanyuan Wang

TL;DR

MicLog tackles accurate and efficient log parsing by applying ProgMeta-ICL to small open LLMs. It integrates a weighted sampling module, progressive meta-training from zero-shot to few-shot, and a multi-level cache to rapidly match logs to templates and curtail LLM queries. On Loghub-2.0, it attains 97.6% parsing accuracy, 95.3% template precision, and 90.5% template recall, while speeding parsing by 42.4% over AdaParser. The work demonstrates that progressive meta-learning with open models can achieve robust cross-domain log parsing with privacy-preserving, cost-efficient deployments.

Abstract

Log parsing converts semi-structured logs into structured templates, forming a critical foundation for downstream analysis. Traditional syntax and semantic-based parsers often struggle with semantic variations in evolving logs and data scarcity stemming from their limited domain coverage. Recent large language model (LLM)-based parsers leverage in-context learning (ICL) to extract semantics from examples, demonstrating superior accuracy. However, LLM-based parsers face two main challenges: 1) underutilization of ICL capabilities, particularly in dynamic example selection and cross-domain generalization, leading to inconsistent performance; 2) time-consuming and costly LLM querying. To address these challenges, we present MicLog, the first progressive meta in-context learning (ProgMeta-ICL) log parsing framework that combines meta-learning with ICL on small open-source LLMs (i.e., Qwen-2.5-3B). Specifically, MicLog: i) enhances LLMs' ICL capability through a zero-shot to k-shot ProgMeta-ICL paradigm, employing weighted DBSCAN candidate sampling and enhanced BM25 demonstration selection; ii) accelerates parsing via a multi-level pre-query cache that dynamically matches and refines recently parsed templates. Evaluated on Loghub-2.0, MicLog achieves 10.3% higher parsing accuracy than the state-of-the-art parser while reducing parsing time by 42.4%.

MicLog: Towards Accurate and Efficient LLM-based Log Parsing via Progressive Meta In-Context Learning

TL;DR

MicLog tackles accurate and efficient log parsing by applying ProgMeta-ICL to small open LLMs. It integrates a weighted sampling module, progressive meta-training from zero-shot to few-shot, and a multi-level cache to rapidly match logs to templates and curtail LLM queries. On Loghub-2.0, it attains 97.6% parsing accuracy, 95.3% template precision, and 90.5% template recall, while speeding parsing by 42.4% over AdaParser. The work demonstrates that progressive meta-learning with open models can achieve robust cross-domain log parsing with privacy-preserving, cost-efficient deployments.

Abstract

Log parsing converts semi-structured logs into structured templates, forming a critical foundation for downstream analysis. Traditional syntax and semantic-based parsers often struggle with semantic variations in evolving logs and data scarcity stemming from their limited domain coverage. Recent large language model (LLM)-based parsers leverage in-context learning (ICL) to extract semantics from examples, demonstrating superior accuracy. However, LLM-based parsers face two main challenges: 1) underutilization of ICL capabilities, particularly in dynamic example selection and cross-domain generalization, leading to inconsistent performance; 2) time-consuming and costly LLM querying. To address these challenges, we present MicLog, the first progressive meta in-context learning (ProgMeta-ICL) log parsing framework that combines meta-learning with ICL on small open-source LLMs (i.e., Qwen-2.5-3B). Specifically, MicLog: i) enhances LLMs' ICL capability through a zero-shot to k-shot ProgMeta-ICL paradigm, employing weighted DBSCAN candidate sampling and enhanced BM25 demonstration selection; ii) accelerates parsing via a multi-level pre-query cache that dynamically matches and refines recently parsed templates. Evaluated on Loghub-2.0, MicLog achieves 10.3% higher parsing accuracy than the state-of-the-art parser while reducing parsing time by 42.4%.
Paper Structure (22 sections, 7 equations, 4 figures, 4 tables, 1 algorithm)

This paper contains 22 sections, 7 equations, 4 figures, 4 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Log Parsing
  4. Large Language Models and In-Context Learning
  5. Meta In-Context Learning
  6. Method
  7. Weighted DBSCAN Sampler
  8. Progressive Meta In-Context Training
  9. Multi-level Cache Enhanced LLM ICL Log Parser
  10. Evaluation
  11. Experiment Setup
  12. Datasets.
  13. Baselines.
  14. Metrics.
  15. Environment and Implementation.
  16. ...and 7 more sections

Figures (4)

  • Figure 1: A simple process of Log Parsing.
  • Figure 2: The workflow of MicLog framework. Sampling and ProgMeta-ICL are performed before online log parsing. The ProgMeta-ICL dataset and inference dataset need to be labeled before progressive meta in-context training and log parsing.
  • Figure 3: Robustness comparison between baselines and MicLog on public datasets (%)
  • Figure 4: Efficiency of MicLog and baselines on Loghub-2.0