Figure 1.
An example of context (pretext/postext) and text extraction.
The size of the pretext/postext used, and the range of text sizes stored, may vary by implementation.
Figure 2.
Tests of identification using full–length queries.
Frequency of success, with 95% confidence intervals, for tests of (A) genus–level identification; (B) weak tests of species–level identification (i.e. those for which no congeners are represented in the data set); (C) strong tests of species–level identification (i.e. those for which congeners are represented in the data set); and (D) all tests of species–level identification. B = BRONX; C = CAOS; D = DNA–BAR/degenbar; F = forced (constrained) tree–search; J = SAP neighbor joining; L = pairwise matching (local alignment); N = NCBI-BLAST; P = pairwise matching (global alignment); S = SAP Barcoder; T = de novo tree–search; and W = WU-BLAST.
Figure 3.
Tests of identification using mini–barcode queries.
Frequency of success, with 95% confidence intervals for tests of (A) genus–level identification; (B) weak tests of species–level identification (i.e. those for which no congeners are represented in the data set); (C) strong tests of species–level identification (i.e. those for which congeners are represented in the data set); and (D) all tests of species–level identification. B = BRONX; D = DNA–BAR/degenbar with redundancy of 10; D = DNA–BAR/degenbar with redundancy of 30; F = forced (constrained) tree–search; J = SAP neighbor joining; L = pairwise matching (local alignment); N = NCBI-BLAST; P = pairwise matching (global alignment); S = SAP Barcoder; T = de novo tree–search; and W = WU-BLAST.
Table 1.
Multiple comparison tests of SIDE genus– and species–level identification performance (p = 0.05).
Table 2.
Similarity of SIDE performance measured by Fleiss' index of interrater agreement ().