1. Background

When a Windows system has been operated as a server over a long period, unexpected errors may occur.  
Windows provides an event log collection function through the *Event Viewer* application.  
By analyzing accumulated historical event logs from multiple perspectives, it is possible to detect early signs of hardware failures and other issues.

 

2. Necessity

When analyzing event logs accumulated over time, the procedure typically follows the steps below, placing a significant burden on system administrators:

1)  Open Event Viewer → Select *Windows Logs* → *System*
2)  Repeatedly perform the following:
    *   Visually inspect each log entry on the screen
    *   Examine messages for entries with high error levels
    *   Record the errors found
3)  Finally, review and analyze the recorded error logs.

If steps 2 and 3 could be handled through natural-language querying that performs trend analysis on the event logs, the workload on administrators could be greatly reduced.

 

3. Adding Event Log Analysis Functions to the RAG Service

​

4. Example Operation

Launch the GUI application within the suite of automated processing tools, enter a prompt into the query window, and the response window will display the output generated by the GPT‑4o‑mini LLM model.
 

​