How to set metadata by using OCR while check in the scanned document?

No, that option is only suitable for scanning of multiple documents with the same layout - typically for a specific form.

You may want to have a look at the M-Files Information Extractor module. If you want to recognize objects such as persons or companies already listed in a value list or an object type in your vault, you can use M-Files Matcher to suggest matches as metadata. If you are looking for other data that can be recognized by a pattern in your document, you can use M-Files Text Analytics to suggest such text strings, dates, mail addresses etc. as metadata. You should be aware though, that OCR is not 100% reliable in recognizing text from a scanned image. It would be much more reliable if you can import the source as a machine readable PDF or other text based document format.