Result on the average precision, recall, and F-measure using our features and domain composition kernel in the best average F-measure case for each set of features.

<p>As sets of features, (F1–5), (F1–6), (F1–5,7), and (F1–7) shown in <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0065265#pone-0065265-t001" target="_blank">Table 1</a> were used. The results by the naive Bayes-based method <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0065265#pone.0065265-Maruyama1" target="_blank">[11]</a>, MCL <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0065265#pone.0065265-Enright1" target="_blank">[7]</a>, MCODE [28], RRW <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0065265#pone.0065265-Macropol1" target="_blank">[18]</a>, and NWE <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0065265#pone.0065265-Maruyama2" target="_blank">[19]</a> are also shown, where the experiments for these methods were performed by <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0065265#pone.0065265-Maruyama1" target="_blank">[11]</a>. (B1), (B2:CC),, (B6) indicate the features by <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0065265#pone.0065265-Maruyama1" target="_blank">[11]</a> (shown also in <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0065265#pone-0065265-t003" target="_blank">Table 3</a>).