When KA crawls/indexes the content attachments, a title for the document is generated from the document.  There are three types of attachments and they are done differently.

 

NOTE:  In collection setup the title of articles and documents can be updated.

 

1.  Microsoft documents 

 

When KA crawls/indexes the Microsoft attachments (like *.doc, *.docx, *.xls, *.xlsx, ...), it will first consider its Title properties (metadata commonly defined in Word, Excel... documents) for for the title of the document.  If the title property is not used then the file name is used.

 

2. PDF documents.

This is described in the documentation here.

 

For PDFs first a title is determined from the text.  If that is not possible then the file Title property is used and lastly the file name is used.

 

3. Text files and other miscellaneous files would just be assigned the file name as the title.