You are here

A new approach for the automated extraction of 3D models from textual descriptions

Martin Unhold (FH Mainz)

GIS or CAD systems are able to handle precise coordinates (x,y,z). Historical texts with descriptions of rooms or buildings do not use them, but other concepts, such as "Object A is on the opposite wall of object B", "Object C has a height of 2m.", "Object D is in the same room as object E", etc. To compute all of these concepts, a space describing ontology (e.g. RCC8-calculus extended) will be created. After that, for every text-to-3d process

  1. Identify expressions in the text, that describe spatial objects (e.g. altar, window) and spatial relations (e.g. behind, inside).
  2. Transform these expressions into the space describing ontology.
  3. Transform the out coming semantic graph into a 3D-model.
13.03.2014 - 09:15
B 017