combinFormation: Mixed-initiative composition of image and text surrogates promotes information discovery