Table 2.

Overview of similarity measures used in research articles.

Similarity measures	Description	Refs.
logD	logD ratio (I, II) = logDI/logDII	[37]
Chemical nature	Compounds were clustered based on their chemical nature, i.e., acids, bases and neutrals.	[54]
k	k-ratio similarity = kI/kII	[55]
Dual filtering	Dual filtering involves using a combination of two similarity searches to improve the model performance, such as combinations of structural similarity searching and retention time similarity searching, combining partition coefficient logP searching with SDI-based searching.	[56]
Tanimoto similarity	Tanimoto similarity (I, II) = $\frac{{\text{FP}}_{Ⅰ}·{\text{FP}}_{Ⅱ}}{{\Vert {\text{FP}}_{Ⅰ}\Vert }^2+{\Vert {\text{FP}}_{Ⅱ}\Vert }^2-{\text{FP}}_{Ⅰ}·{\text{FP}}_{Ⅱ}}$	[57]
SDI	logα = log (k/kEB) = η′H – σ′S + β′A + α′B + κ′C	[58]

I, II: different compounds; FP: fingerprint; logD: distribution coefficient (logarithmic scale); α: chromatographic selectivity; k: retention factor; k_EB: the retention factor of ethylbenzene; SDI: second dominant interaction; H, S, A, B, C: five column coefficients; η′, σ′, β′, α′, κ': five complementary solute coefficients.