Figure 5.

Statistical significance with respect to the average size of modules. The average -log(p-value) of modules with respect to their average size in each output set. Four distinct methods were implemented: the flow-based algorithm using semantic similarity and semantic interactivity, the betweenness-cut algorithm and the betweenness cut algorithm with the pre-processing step to filter out the edges with low semantic similarity. The betweenness-cut algorithm was enhanced by the pre-process, and the flow-based algorithm outperformed the betweenness-cut algorithm.