ABSTRACT
The categories generated via taxonomies lead directly to connections through the use of orthogonal comparison. ese connections are oen evidence of some fundamental relationship that oen enables signicant insights about the organization contained within the information space. A high degree of co-occurrence is the rst symptom of a connection. Co-occurrence means that two dierent categories or entities occur together. To nd the best connections, a proper statistical analysis should be employed, that is, one that does not simply look at the size of a particular co-occurrence but also takes the context of that co-occurrence into account. Assuming no relationship exists that ties together two particular categories from dierent taxonomies, we would expect to nd around (A × B) occurrences of A and B together, where A is the percentage of times one category occurs and B is the percentage of times the other occurs. Let us call this expected value E. An exceptional value means that A × B > E, indicating an association that is stronger than we expected. e value of E will usually will be dierent for every combination of categories, since the size of each category may be dierent.
