This is the current expected behavior, so I'm creating this ticket as 'Suggestion'.

Summary

When updating the attribute values that refer to the other objects via Assets CSV Import, the existing attribute values are not cleared even if all the values in the CSV don't match any other objects.

How to replicate

1. Create an object type named "Department" with the following attributes
   • Department ID (label attribute)
   • Name
2. Create an object type named "Employee" with the following attributes
   • EmpID (label attribute)
   • Name
   • Department (referencing the objects in the object type "Department")
3. Create two objects under Department

   • Department ID	Name
     1000	Sales
     1001	Support

1. Create an object under Employee via CSV import
   • Set the reference mapping below on the attribute mapping setting in the import configuration:

     "objecttype" = "Department" AND "Department ID" IN (${Department${0}})
     

• • Set Missing objects = Ignore and Missing objects outbound reference = Remove (or Missing objects = Remove; the result will be same)
  • Sample CSV:

    EmpID,Name,Department
    N1000,Test User,1000
    

1. Confirm that the import succeeds, and the employee is created with the department value "Sales"
2. Using the same import configuration as in step 4, update the existing object with a Department ID that does not exist.
   • Sample CSV:

     EmpID,Name,Department
     N1000,Test User,1002
     

Expected Results:

The object with EmpID = N1000 doesn't have any Department value after the import because there are no objects with Department ID = 1002.

The following search should be performed during the import as per the reference mapping.

"objecttype" = "Department" AND "Department ID" IN (1002)

Actual Results:

The object with EmpID = N1000 still has the value "Sales" in the Department attribute after the import.