Figure 3

Conceptual illustration of how an inferred association between two objects is identified and evaluated using the methods in Equations 6–7. A primary object of interest (A – black node) co-occurs in records with other objects (B – gray nodes). Each association (A-B and B-C) is assigned a mutual information score, the higher the score the stronger the association between the two. These intermediate associations (B) can then be used to infer an association between A and C (white nodes). Example values are given above the lines. Note that, depending upon whether these values are averaged or the minimum value taken, the rank order of the implicit connections changes. If averaging, the bottom C node receives a higher score (11.5 vs. 9). If taking the minimum, the top node receives a higher score (8 vs. 6).