engleski

CroNER: Recognizing Named Entities in Croatian Using Conditional Random Fields

In this paper we present CroNER, a named entity recognition and classification system for Croatian lan-guage based on supervised sequence labeling with conditional random fields (CRF). We use a rich set of lexical and gazetteer-based features and different methods for enforcing document-level label consistency. Extensive evaluation shows that our method achieves state-of-the-art results (MUC F1 90.73%, Exact F1 87.42%) when compared to existing NERC systems for Croatian and other Slavic languages.

named entity recognition ; conditional random fields ; natural language processing ; information extraction ; Croatian language

nije evidentirano