Аннотация
NER have been regarded as an efficient strategy to extract relevant entities for various purposes. The aim of this paper is to exploit conventional method for NER in Odia by parameterizing CRF++ tool in different ways. As a case study, we have used gazetteer and POS tag to generate different feature set in order to compare the performance of NER task. Comparison study demonstrates how proposed NER system works on different feature set.
Пользователи данного ресурса
Пожалуйста,
войдите в систему, чтобы принять участие в дискуссии (добавить собственные рецензию, или комментарий)