WOW! That's a big answer!
Basically, it starts with the entity relationship so in the knowledge graph it used real world data to determine that could mean those two different things.
It then looks inside the content for the context and the meaning inside, so if the page was talking about flowers, viola means flowers because in the knowledge graph, violas and flowers are very close together mathematically with a very strong relationship.
If you was talking about instruments, it would know it would know that.
If you was talking about instruments and flowers on the same page, Google might have a difficult time what you meant and that is where you might wind up with those, well it wouldn't give RankBrain to you but if you were searching for it they would be like - what's viola?
If you are not specific in what you are searching, Google would pull back RankBrain from within the search result if it wasn't sure what relationship was there.
Answered by: Kristine Schachinger