11 novembre 2016
https://creativecommons.org/licenses/by-nc-nd/4.0/ , info:eu-repo/semantics/openAccess
Michele Filannino et al., « Gold standard vs. silver standard: the case of dependency parsing for Italian », Accademia University Press, ID : 10.4000/books.aaccademia.1475
Collecting and manually annotating gold standards in NLP has become so expensive that in the last years the question of whether we can satisfactorily replace them with automatically annotated data (silver standards) is arising more and more interest. We focus on the case of dependency parsing for Italian and we investigate whether such strategy is convenient and to what extent. Our experiments, conducted on very large sizes of silver data, show that quantity does not win over quality.