Objects in the image can be members of
multiple parsed structures

This parsing example is based on a four-part x,y data graph. When the parsed elements for the upper-right data graph, "Site 3", are highlighted, we can see that various components below and to the left, but quite distant from the data itself, are part of the solution. The y-axis label on the far left is shared by all four of the parses for "Site 1" through "Site 4, that is, it appears as a constituent in all four of these components of the overall image parse. The numerical labels and tick marks on the upper left are shared by "Site 1" and "Site 3", while the numerical labels on the lower right are shared by "Site 3" and "Site 4". The x-axis label at the very bottom is shared by all four data graph parse components. Sharing is common in graphics, but is not allowed in many approaches to parsing. The "remote" components that end up being shared are found by using X-Or rules that first look for nearby elements, but if they cannot be found, remote but properly aligned elements are searched for (see the X-or construct in the X-Ticks rule in the complete x,y grammar).


Next: Spatial indexing and geometrical computations

Back to Index page