As current search technology shows, just searching for the query terms does not suffice. For delivering better search results semantically equivalent and similar terms need to be considered too. We argue that in order to build helpful search functions, the background knowledge represented for semantic search should represent what the average user considers to be equivalent or similar.