4 - annotation (Van Rooy)

what is corpus annotation according to Van Rooy?

Corpus annotation is the practice of adding interpretative linguistic information to an electronic corpus of spoken and/or written language data

Which types of annotation are there?

• automatic annotation

• semi-automatic annotation

• manual annotation

Which types of automatic annotation are there?

• POS-tagging

• syntactic parsing

• semantic tagging

What is POS-tagging?

POS-tagging is an annotation method in which each word in the corpus is ‘tagged’ (labelled) with information about its grammatical category.

What do CLAWS tags mean?

It is a categorization for POS-tagging made by Lnacaster University, and stands for Constituent Likelihood Automatic Word-tagging System.

How reliable are automatic annotation tools?

They don’t perform with 100% accuracy, making it necessary to check for tool performance, either through precision, recall, or f-score.

What is precision, recall, and f-score?

They are ways of checking the performance of an automatic annotator tool.

• precision checks whether the label given by a POS tagger, is actually correct. So in case of an item identified as a phrasal verb, checking whether that item is, in fact, a phrasal verb.

• recall checks whether the POS tagger missed any items for that label, so did the POS tagger miss any phrasal verbs?

• F-score combines both precision and recall into one value, it represents the balance between precision, and recall.