The function of system monitoring is to track the running status of the knowledge base in real time (such as visits, query success rate, content update) and abnormalities (such as failures, inefficient retrieval) to ensure its stable and efficient operation. Applicable scenarios are scenarios that need to ensure the continuous availability of the knowledge base and optimize the operation efficiency. In particular, it is suitable for high-traffic customer service knowledge base operation and maintenance, enterprise core business knowledge system stability assurance, and multi-end point knowledge base service quality monitoring. This article describes how to monitor assets and models.

1. Go to the Enterprise Knowledge Engine console and click Project Center in the top navigation bar.

2. Select System Monitoring > Asset Monitoring in the left navigation bar.
   

3. On the asset monitoring page, set a time interval to filter out information such as the number of AI applications associated with the enterprise knowledge engine, content usage effectiveness (such as recall times), and so on.

4. Click on the update record to view the update record of the enterprise knowledge engine, which can be filtered by parameters such as operation type and operation object. The update record of the same knowledge base stores the latest 3 months of data by default.
   

   Parameter	Explain
   Number of associated AI applications	The total number of AI applications currently tied to the enterprise's knowledge engine
   Recall times	Total number of times enterprise knowledge engine content is invoked by AI applications
   Number of sessions	Total number of sessions using AI apps
   Number of likes/clicks	During the AI application session, the total number of likes and clicks on the reply question by the user
   Operation type	Radio selection, including create, copy, edit, import, delete, re-sharding. Default "All operation types" warning When filtering, merge types, edit knowledge base names, edit knowledge base descriptions, and edit paragraphs are all edits
   Action object	Single selection, including knowledge base, documentation, sharding. Default "All Object Types"
   Update Person	Radio selection, default "All Updater"
   Update time	Default last 30 days

1. Go to the Enterprise Knowledge Engine console and click Project Center in the top navigation bar.

2. Select System Monitoring > Model Monitoring in the left navigation bar.

3. On the model monitoring page, you can filter and view the called model performance indicators (such as call amount, Token consumption, response speed, etc.) by selecting the user, application, and time, providing data support for resource optimization and exception troubleshooting.
   

   Parameter	Explain
   Total number of calls	Associated model, total number of calls
   Total Number of Tokens Invoked	Input and output Token cumulative consumption
   Enter Total Tokens	Input Token Cumulative Consumption
   Number of Tokens Output	Cumulative consumption of output Tokens
   Average response duration	Time taken from request initiation to return result (milliseconds)