elasticsearch-skroutz-greekstemmer
elasticsearch-skroutz-greekstemmer copied to clipboard
rule0, exceptions handling
rule0
of SkroutzGreekStemmer.java
tries to handle special cases for specific word endings.
However, most of those cases concern whole words, rather than endings.
Eg. the word περατοσ
is handled as an ending, and will also match υδατοπερατοσ
and stem it as υδατοπερ
, σαφωσ
will match φωσ
, etc.
Those case are false positive matches.
Most of the cases should be handled with string equality (rather than string suffix matching).
This should happen in an extra step before what now is rule0
and rule0
should have less special cases to handle