Create a token filter or a tokenizer or even an option that removes duplicate words in fields  to avoid misleading scoring. 
For example: if I insert a document with a property "Testing testing document", another with "Testing example" and a third with "Testing foo" and I search for "Testing foo", the first document receives a higher score.