Figure 3.

Statistics of the corpus. In the whole corpus, 5 dominate detection methods take up nearly 59.3% occurrences and 86.1% (99 out of 115) methods occur in less than 10% documents.