Seminář: Using Conditional Random Fields for Named Entity Recognition in Medical Texts, European Semantic Web Conference (ESWC 2006), Budva

Presentation of work done during an internship at National Institute of Informatics, Japan. Conditional Random Fields were used to solve Named Entity Recognition task on information about medical drugs. The information was available in Czech, therefore morphological tools specialized for Czech language had to be used for preprocessing. A combination of morphological and semantic features was then used as input for machine learning algorithm. The implementation of CRF used in experiments was CRF++.