Data
Human Alignment
Available for the training set and the
testing set.
Each sentence pair is separated by an empty line. For each pair, the first
sentence is from the Elementary version of the article, and the second is from
the Comprehensive version of the article. The number at the beginning of each
sentence is the sentence number in the article.
The instructions for the annotation are available here.
Automatic Alignment
Available only for the testing set. The format is the
same as for the human alignments. (These alignments were produced by varying
the parameters of the system, for a recall of 55.8%).
Alignment Plots
For the training set the human alignment plots are available here. For the testing set the plots for both the human and the automatic alignments are available here.