Publication at Faculty of Arts |
2015
Abstract
In this study, we aim to verify the reliability of the annotation of idioms in spoken corpora. Idioms are searched for and annotated using a special tool.
Some Czech idioms come in different lengths, word order permutations and variants. These properties greatly complicate their identification.
Somatic idioms are among the most common idioms in language. They can be easily retrieved by keyword (the name of the part of the human body).
They are suitable for verifying the accuracy of annotation. For the evaluation, we use the well-known precision and recall measures.