Abstract: A system for working with data in datasets or information structures, wherein at least some of the data has latent preferences for connection with the data in other datasets or information structures. The data for each of the datasets or information structures is stored into a slice of separable cells having actual connections forming independent relational orderings in one or more dimensions. The cells in each slice are then examined for the latent preferences and the cells having data seeking connection is identified as choosing cells and the cells sought to be connected to are identified as target cells. A latent connection is then added to either the choosing cell or the target cell for each of the latent preferences, thus preparing the slices for potential expression into one or more conglomerate structures.