Automatically generated NE tagged corpora for English and Hungarian
Eszter Simon, Dávid Márk Nemeskey · SZTAKI Publication Repository (Hungarian Academy of Sciences) · 2012
Supervised Named Entity Recognizers require large amounts of annotated text. Since manual annotation is a highly costly procedure, reducing the annotation cost is essential. We present a fully automatic method to build NE annotated corpora from Wikipedia. In contrast to recent work, we apply a new method, which maps the DBpedia classes into CoNLL NE types. Since our method is mainly language-independent, we used it to generate corpora for English and Hungarian. The corpora are freely available.