This paper presents an overview of work based on the QVIZ-project to support faceted browsing, focusing on the handling of larger, more complex relational database structures, discrete and continuous data, hierarchies, temporal and spatial data. Faceted browsing allows the creation of unpredictable arrangements of search criteria by the user. Such dynamics require a generic and abstracted mechanism in order to be able to adapt to multidimensional exploration and user requirements. Faceted browsers function through the progressive narrowing of choices in selected dimensions. This paper describes an approach using a graph representation of data models and shortest path operations to build queries. The system described is fully functional and has developed since 2007 at HUMlab, Umeå University, Sweden. It is now being used in several digital humanities and multidisciplinary projects with different database schemata.