81  Last view: 2022-03-04

English/Danish/Spanish regular polysemy corpus

The result of an annotation task on regular polysemy for a series of semantic classes or dot types in English, Danish and Spanish.
Each of the data files has 500 examples with human annotations for examples like :
"Manuel died in exile in 1932 in <England>", where the highlited word England is potentially metonymic.
The highlighted words can be literal, metonymic or underspecified.

The final sense tag is calculated in two different ways from the N annotations each example has:
a) a voting scheme with a back off that assigns the underspecified sense when literal and metonymic tie.
b) an unsupervised method that assigns weights to annotators (MACE, Hovy et al 2013)

The nine datasets are
English:
1.- Animal/Meat
2.- Artifact/Information
3.- Container/Content
4.- Location/Organization
5.- Process/Result

Danish
6.- Container/Content
7.- Location/Organization

Spanish
8.- Container/Content
9.- Location/Organization