Modifications from v1.0.1:
- MATLAB implementation now as function with input
<config.file.json>
- normalization dataset is now a parameter in config file:
"WEIGHTS_NAME"
- fixed bug when dealing with very sparse sub-matrices. Such a matrix can result in
nan
values inS
; these are ignored when computing similarity