Spatial Similarity Metrics

Com­par­ing space allo­ca­tion pro­grams com­pu­ta­tion­al­ly is a resource inten­sive task. This paper intro­duces a method which reduces the com­plex­i­ty of floor plan graphs to facil­i­tate the prob­lem. In a first step the nodes are labeled accord­ing to a laid out clas­si­fi­ca­tion scheme. This in its own right reduces the com­plex­i­ty of the com­par­i­son by intro­duc­ing more diver­si­ty among the nodes.

Sub­se­quent­ly the graph can be coars­ened in sev­er­al steps based on the cat­e­go­ry infor­ma­tion. This does not only fur­ther reduce the nec­es­sary com­pu­ta­tion, but can also offer a visu­al­iza­tion of and addi­tion­al insight into the essen­tial struc­ture of the space allo­ca­tion pro­gram at hand.