The 4W (What-Where-When-Who)project has demonstrated that the combination of rich content delivering devices, such as smart phones, together wireless broadband networking and server side computational power can be used to seamlessly weave a mesh of smart services and intelligent environments providing support to human memory capabilities. Our system is capable of capturing user needs and to-dos, infer where those needs and tasks can be efficiently fulfilled/performed and provide timely and localized hints about the identified sweet spots. However, to perform this task an efficient way of classifying user provided needs and to-dos is necessary. In this paper we describe how we leveraged a social-network like approach to populate and enrich our classification engine in a user-controlled, cooperative way.