Named Entity Recognition in the Climate Change domain - An examination of NER systems for climatological knowledge discovery

Abstract

This study evaluates existing Named Entity Recognition systems for their viability as a part of the larger Ocean-Certain project. Motivation for the Ocean-Certain initiative and this project is given, together with research goals. Background information spanning Ocean-Certain, Named Entity Recognition, and evaluation methods is provided to give context. A set of Named Entity Recognition systems is then selected for formal evaluation. Finally, each candidate system is tested against a manually annotated gold standard corpus. The results of each evaluation are then analysed and discussed. Possible sources of errors and steps that can be taken to improve each system are then detailed. At the end of this study one system is selected for improvement and testing versus an expanded gold standard.