Figure 1
From: Indirect two-sided relative ranking: a robust similarity measure for gene expression data
Direct vs. Indirect Similarity. (a) The direct similarity method evaluates a given pair of instances without taking into account any other knowledge in the database. (b) The indirect similarity method benefits from the extra knowledge within the database, comparing the similarity of the pair to all the instances in the database and uses this information to determine the similarity between a given pair of instances.