Skip to main content

Table 2 Performance of the automatic sentence extraction step.

From: Corpus Refactoring: a Feasibility Study

Overall performance

Percent

Count

Correct extraction

66%

187/283

Incorrect extraction

33%

96/283

Total

100%

283/283

Type of error

Percent

Count

Too little extracted

48%

46/96

Title text not extracted

39%

37/96

Too much extracted, expanded text selection

9%

9/96

Too much extracted

4%

4/96

Total

100%

96/96