The need of improving the privacy on data publisher becomes more important because data grows very fast. Traditional methods for privacy preserving data publishing cannot prevent privacy leakage. This causes the continuous research to find better methods to prevent privacy leakage. K-anonymity and L-diversity are well-known techniques for data privacy preserving. These techniques cannot prevent the similarity attack on the data privacy because they did not take into consider the semantic relation between the sensitive attributes of the categorical data. In this paper, we proposed an approach to categorical data preservation based on Domain-based of semantic rules to overcome the similarity attacks. The experimental results of the proposal approach focused to categorical data presented. The results showed that the semantic anonymization increases the privacy level with effect data utility.