However, algorithms are not perfect; the missing link prediction can be wrong. That is because real-world scenarios are far more complicated, and it is difficult to consider every nuance in all domains for the algorithm design. An analyst's prior knowledge is needed for further examination of the outputs from the algorithm, which combines the flexibility of humans and the scalability of machines.

Example implementations involve a visual interface to help analysts better make sense of the missing links identified by the aforementioned methods in bipartite networks. This visualization module involves five interactively-coordinated views: a Network View and a Link List View to support the exploration of missing links, a Motifs Overview and a Detail View to offer the analysis of motifs, and a Metrics View to display node-based metrics as illustrated in FIG. 3. These views present the outputs from the analysis module in visual forms to allow analysts to effectively answer the what, why, and how questions about missing links.