A privacy-preserving distributed filtering framework for NLP artifacts

Table 1 Identification of globally infrequent bigrams

Data Owner	Frequency of the bigram Flu-fever	Frequency of the bigram Cancer-pain	Frequency of the bigram Diabetes-glaucoma
A	10	15	20
B	20	15	10
C	5	15	25
Total	35	45	55

Let us consider the data of the above table. Assume, the threshold value is 40. Since total count of Flu-fever (35) is less than the threshold value (40), it will not be considered privacy-sensitive

ISSN: 1472-6947