Skip to main content
. 2006;2006:599–603.

Table 2.

Sentence-level evaluation of the summarization at different compression ratios with different feature sets. Best results for each compression ratio are shown in boldface. P = precision; R = recall; F = F-score.

Compression Ratio
0.1
0.2
0.3
0.4
P R F P R F P R F P R F
Baseline .25 .11 .15 .25 .20 .22 .25 .31 .27 .25 .40 .30
MMR .44 .19 .27 .38 .31 .34 .36 .44 .39 .34 .57 .42
(1) MMR+pos+num+len .44 .19 .26 .40 .33 .36 .38 .48 .42 .36 .58 .44
(1)+polarity (auto) .45 .20 .27 .42 .35 .38 .39 .49 .43 .37 .61 .46
(1)+polarity (manual) .49 .21 .29 .44 .38 .40 .41 .52 .46 .38 .64 .48
(1)+outcome (auto) .45 .20 .27 .41 .35 .38 .39 .48 .43 .37 .61 .46
(1)+outcome (manual) .51 .22 .31 .45 .38 .41 .42 .53 .46 .39 .65 .48