Currently as a any Doc instance initializes, the sentencer known as sentence boundary detector works to split every raw document into sentences and for sadedegel.Sentences object. This causes a slowdown in any process that does not require sentence splitting beforehand.
Proposed solution is to make sbd work only when .sents attribute is invoked by the user or any other user implemented/overriden(__iter__, __len__) methods.
askarbozcan
commented
2 years ago
Currently as a any
Docinstance initializes, the sentencer known assentence boundary detectorworks to split every raw document into sentences and forsadedegel.Sentencesobject. This causes a slowdown in any process that does not require sentence splitting beforehand.Proposed solution is to make
sbdwork only when.sentsattribute is invoked by the user or any other user implemented/overriden(__iter__,__len__) methods.