Personalized web search and recommendation systems aim to provide results that match the users' personal interests and lead to a more satisfactory and effective information access. Building user profiles that reflect a large spectrum from continuous (long-term) to specific (short-term) interests is an essential task when developing personalized web applications. In this paper we present a method to generate user interest profiles without direct user interaction generated out of data sourced from quiz games played by the user. Both utilized games, WhoKnows? and RISQ! , have originally been developed as serious games with the intention to rank facts in knowledge representations as well as to find inconsistencies in a given knowledge base.