The process receives a visual specification, which specifies data sources, visual variables, and data fields from the data sources. Each visual variable is associated with one or more data fields and each data field is either a dimension or a measure. For each measure m, the process identifies a set R(m) consisting of dimensions that are reachable from the measure by a sequence of many-to-one relationships in an object model for the data sources. For each distinct set R, the process forms a data field set S, consisting of each dimension in R and each measure m for which R(m)=R. For each set S and for each measure in the set S, the process aggregates values of the measure according to the dimensions in S. The process builds a data visualization according to the data fields in the set S and the visual variables they are associated with.