Fig. 4

Anonymizing Seer Data by using Pig Latin with four Q-ID attributes. This figure represents our MDSBA method structure. It depends on iterating processes of grouping data. If a group of four Q-ID attributes are chosen. Stages of grouping by four Q-IDs, then by three Q-IDs, then by two Q-IDs and finally by one Q-ID. The grouping chooses the highest three Q-IDs probability, then highest two Q-IDs probability, and finally the highest one probability. In each stage, the groups are successful when the number of equivalent or semi-equivalent records is greater than \(\overline{k}\). The records < \(\overline{k}\) are ungrouped, and then regrouped again