Table 5 Results of the Shapiro-Wilk test for normality for each metric and prompt. The W statistic measures how well the data follows a normal distribution, with a lower p-value indicating a departure from normality.
From: Large scale summarization using ensemble prompts and in context learning approaches
Metric | Prompt | W | p-value |
|---|---|---|---|
SSF | Vanilla | 0.9791 | 0.05279 |
CoD | 0.9790 | 0.05243 | |
Ev2 | 0.9548 | 0.0579 | |
RLF | Vanilla | 0.7753 | \(3.04 \times 10^{-7}\) |
CoD | 0.5440 | \(4.13 \times 10^{-11}\) | |
Ev2 | 0.6992 | \(1.01 \times 10^{-8}\) | |
Cons | Vanilla | 0.7808 | \(3.99 \times 10^{-7}\) |
CoD | 0.5649 | \(7.97 \times 10^{-11}\) | |
Ev2 | 0.6840 | \(5.48 \times 10^{-9}\) | |
RDF-CRTD | Vanilla | 0.8765 | \(1.02 \times 10^{-4}\) |
CoD | 0.6396 | \(1.02 \times 10^{-9}\) | |
Ev2 | 0.9423 | 0.0181 | |
BAA | Vanilla | 0.7064 | \(1.37 \times 10^{-8}\) |
CoD | 0.5542 | \(5.68 \times 10^{-11}\) | |
Ev2 | 0.7420 | \(6.35 \times 10^{-8}\) | |
SUSWIR | Vanilla | 0.9104 | 0.0012 |
CoD | 0.9535 | 0.0513 | |
Ev2 | 0.9088 | 0.0011 | |
NIC | Vanilla | 0.4116 | \(9.68 \times 10^{-13}\) |
CoD | 0.9541 | 0.0544 | |
Ev2 | 0.9370 | 0.0114 |
