Publication at Faculty of Mathematics and Physics |
2009
Abstract
In this paper, we present some organizational aspects of building of a large corpus with rich linguistic annotation, while Prague Czech-English Dependency Treebank (PCEDT) serves as an example. We stress the necessity to divide the annotation process into several well planed phases.
We present a system of automatic checking of the correctness of the annotation and describe several ways to measure and evaluate the annotation and annotators (inter-annotator accord, error rate and performance).