Random connectivity is not be what we observe in the NMAP corpus.
To test this hypothesis, additional corpuses have been randomly generated containing the same number of sets as the original corpus but where each feature is independent yet has the same likelihood of occurrence that is observed in the original corpus from our NMAP scan.
Because of this assumption of independence, structural characteristics in the original corpus should not be found in this generated corpus.
INSERT FREQUENCY DISTRIBUTION OF FEATURES
INSERT TREES GENERATED FROM TABLES BUILD FROM I,SUSHI FROM BOTH CORPUSES
INSERT FREQUENCY CHARACTERISTICS OF SYMBOLS FROM CORPUSES