The goal of this work is to improve the performance of a neural named entity recognition system by adding input features that indicate a word is part of a name included in a gazetteer. This article describes how to generate gazetteers from the Wikidata knowledge graph as well as how to integrate the information into a neural NER system. Experiments reveal that the approach yields performance gains in two distinct languages: a high-resource, word-based language, English and a high-resource, character-based language, Chinese. Experiments were also performed in a low-resource language, Russian on a newly annotated Russian NER corpus from Reddit tagged with four core types and twelve extended types. This article reports a baseline score. It is a longer version of a paper in the 33rd FLAIRS conference (Song et al. 2020).
@article{arxiv.2003.03072,
title = {Improving Neural Named Entity Recognition with Gazetteers},
author = {Chan Hee Song and Dawn Lawrie and Tim Finin and James Mayfield},
journal= {arXiv preprint arXiv:2003.03072},
year = {2020}
}
Comments
Short version accepted to the 33rd FLAIRS conference