LogSay

Pipeline

Script Execution: run.sh

1. Train the Question-Event Matching Model: python q2e_model.py

2. Save embeddings corresponding to each event: python bert_embedding.py

3. Evaluate the results of the Question-Event Matching Model: python Q2E.py

4. Filter original logs based on events matched with each question and save filtered QA results: python QE2Log_model.py (model-based)

5. Train QANet to extract answer positions from events: cd QANet-pytorch-/, then python main.py --mode data and python main.py --mode train

6. Evaluate QANet results and save positions of answers for each question in log events: cd QANet-pytorch-/, then python main.py --mode test

7. Extract answers to questions based on results from steps 4 and 5

8. Determine computations to be performed based on questions

LogSay

Pipline

运行脚本: run.sh

1. 训练问题-事件匹配模型: python q2e_model.py

\

2. 保存每个事件对应的embedding: python bert_embedding.py

3. 评估问题-事件匹配模型的的结果: python Q2E.py

4. 根据每个问题匹配的事件, 过滤原始logs, 保存过滤后的QA结果 python QE2Log_model.py(model-based)

5. 训练QANet提取事件中答案的位置: cd QANet-pytorch-/ python main.py --mode data python main.py --mode train

6. 评估QANet的结果，并保存每个问题对应答案在日志事件中的位置: cd QANet-pytorch-/ python main.py --mode test

7. 根据4、5的结果提取问题的答案

8. 根据问题判断应该进行的计算