The second alignment is composed with the first one meaning that for
any pair (o, o', n, r) in O and (o',o", n', r') in O' the resulting
alignment will contain:
( o, o", join(n,n'), compose(r, r')) iff compose(r,r') exists.
Effort-based relaxed precision and recal similarity
Note: it will be better if the parameters were replaced by the actual sibling (choice)
Table 3 of [Ehrig2005]
Symmetric relaxed precision and recal similarity
This computes similarity depending on structural measures:
the similarity is symALPHA^(val1+val2), symALPHA being lower than 1.
Cut refinement :
- getting those cells with strength above n (hard)
- getting the n best cells (best)
- getting those cells with strength at worse n under the best (span)
- getting the n% best cells (perc)
- getting those cells with strength at worse n% of the best (prop)
- getting all cells until a gap of n (hardgap)
- getting all cells until a gap of n% of the last (propgap)
Rule:
threshold is betweew 1 and 0