A linear classifier based on entity recognition tools and a statistical approach to method extraction in the protein-protein interaction literature

Table 11 The distribution of the secondary labels for sentences tagged as Y by majority of annotators

Label	# of sentences tagged by the Majority as Label	% with respect to all Y-tagged sentences (755)	% with respect to all sentences (1049)
Y2	199	26%	19%
Y1	172	23%	16%
Y0	297	39%	28%

Annotators assigning a "Y" to a sentence were further asked to assign a numeric label, indicating the actual protein-protein interaction content of the sentence, as follows: 2 - If Protein-protein interaction (PPI) is directly and explicitly mentioned within the sentence (along with the method of detection); 1 - if PPI is implied in the sentence (along with the method of detection), but not explicitly stated; 0 - if PPI is neither implied nor mentioned in the sentence.
The table shows the number of sentences labelled as Y2, Y1 and Y0 by a majority of the annotators, as well as the percentage with respect to the total number of sentences labelled as Y, and with respect to the whole collection of labelled sentences.
Note that the total number of majority Y2, Y1 and Y0 labels in the second column on the left does not sum to 755 (and the respective percentages do not sum to 100%), as for some of the sentences in which two or more annotators agree on the "Y" tag, there is not necessarily such agreement on the additional numerical label (0, 1 or 2).

ISSN: 1471-2105