[developers] processing of lexical rules

[Berthold Crysmann]

Hi Stephan,

what is the current situation for performance tuning of morphological 
rules in the LKB?

Are there any features other than limiting the maximal depth of rule 
application?

It appears that sequences of highly ambiguous affixes take a lot of 
time. Segmentation is quite speedy, but replay can be painfully slow. 
Does the rule filter not apply?

Any ideas?

Berthold