EYDES  EYDES
 

Enhanced Access (5):

Visualization of a Rule Based Algorithm

detected areal patterns 
the principle

phenomenon no. 2 : area of more than 40 %
 

Notes:

The task is to find utterances with an environment in which the utterances occur with a percentage of more than 40%. You have to capture the environments for which this is true. 
In this chart the detection of areal structure is realized by a statistical evaluation based on mask-procedures, a variant of pattern matching.
I will demonstrate this technique in a simplified manner - but basically this is the way in which the map above was constructed:
Given a language area with equi-distant representation. Given also a linguistic classification of the representations. We define a square with a margin size of 3 units. 
Now we match the language area for a particular classification: We evaluate every match: if the result of the match is in the defined range we copy the successful area into a derived area. If the result is not successful we copy the area as a blank area into the derived area.
You see the result.
Impact: the open parametrization of the pattern (form and size, exclusive or inclusive, area building)
Problems: marginal areas

Vorherige Folie Nächste Folie Zurück zur ersten Folie Graphik-Version anzeigen