### costsensitive-data

This repository contains data for our EACL 2014 paper [1], in
particular, the doubly-annotated 500 tweets, the new Lowlands Twitter
test set, as well as script to generate the confusion matrix and plot
the Hinton diagram.

If you use any of this, please cite our paper [1].

#### Content: 

	doubly-annotated Twitter data (500 tweets; 100 initial +400 group) 

	Twitter test set created in the Lowlands project [2]
	create confusion matrix of doubly-annotated file and plot Hinton diagram

#### References: 

[1] Barbara Plank, Dirk Hovy and Anders S√łgaard. Learning POS taggers
    with inter-annotator loss. In EACL 2014, Gothenburg, Sweden.

[2] Lowlands project website,

February 13, 2014, B.Plank