Table 4 Quantitative measures to compute the relationship degree between available rating scores and other labelling
From: Text-rating review discrepancy (TRRD): an integrative review and implications for research
Measure | Purpose |
|---|---|
Similarity and Consistency | To determine whether available rating scores correlate with other labelling techniques (lexicon and expert annotators labelling) Example: cosine similarity, Pearson’s correlation coefficient, Spearman’s rho, etc. |
Agreement | To determine whether available rating scores agrees with other labelling techniques (lexicon and expert annotators labelling) Example: per cent agreement, alpha agreement, etc. |
Linear Regression | To determine whether a linear relationship exists between available rating scores and other labelling techniques (lexicon and expert annotators labelling) |