Skip to main content

Advertisement

Table 4 Comparison of different experimental conditions for bm25 and the Lucene ranking algorithm.

From: Is searching full text more effective than searching abstracts?

Model Metric Comparison
bm25 MAP Span (max) + Article, Span (max) >> Abstract, Article
  P20 Span (max) + Article, Span (max), Abstract >> Article
  IP@R50 Span (max) + Article >> Abstract, Article; Span (max) >> Abstract
Lucene MAP Span (max) + Article >> Span (max), Article >> Abstract
  P20 Span (max) + Article > Span (max), Article; Article > Abstract
  IP@R50 Span (max) + Article > Span (max), Article >> Abstract
  1. A >> B indicates that A is significantly better than B (p < 0.01); A > B indicates that A is significantly better than B (p < 0.05);