Semlog

Self-Supervised Log Parsing Using Semantic Contribution Difference

Accepted by Journal of Systems and Software

The code will be reorganized in a more formal manner.

Reproduction

1.pip install -r requirements.txt

2.Run "Semlog/benchmark/Semlog_benchmark.py" to get the results of Semlog directly.

3.Run "logparsers/*.py" to reproduce results of existing parsers

Docker image:

1. docker pull docker.io/gaiusyu/semlog:v1
2. docker run -it --gpus '"device=0"' --name semlog gaiusyu/semlog:v1

The existing parser code reproduced in this paper relies on LogPai.

For the parsers compared in our experiment, we reproduce the code in https://github.com/logpai/logparser.

paper link: https://arxiv.org/pdf/1811.03509.pdf.

Experimental results