Research on Natural Language Text Coreference Resolution Evaluation Algorithms
Dongsheng Liu · 2008
Coreference resolution is a basal and important task in Nature Language Processing (NLP),its availability can give strongly the support for many language engineering applications.Meanwhile,performance evaluation methods and algorithms are crucial in coreference resolution tasks.This paper discusses in detail several current evaluation methods and algorithms,and analyzes their merits and shortcomings,points out that CEAF is better than other algorithms in Chinese natural language text coreference resolution processing under the situation that there are not mass open and sharable annotated corpus focusing on Chinese coreference relation recognition.