In these days, information sharing as a crucial part appears in our vision, bringing about a bulk of discussions about methods and techniques of privacy preserving for data mining which are regarded as strong guarantee to avoid information disclosure and protect individuals' privacy, k-anonymity has been proposed for privacy preserving for data mining and publishing which can prevent linkage attacks by the means of anonymity operation such as generalization and suppression. Manyanonymity algorithms have been utilized for achieving k-anonymity. Here, there is need to discover the relationships between the quasi identifier and other attributes that lead to disclosure the sensitive information. The main goal of this study is to discover the attributes with high variance which lead to disclosure the sensitive information to apply anonymity method on them. While the attributes with low variance, they can consider as quasi-identifier. This study proposed a technique based on transfer the conceptualization of the data base table into another domain which maintains the privacy and reduces the loss of information by decomposing the table using the Singular Value Decomposition (SVD) and revealing latent semantic relationships among attributes in the semantic space using Latent Semantic Analysis (LSA). This technique is the innovative in term of preventing more smart attack which tries to build linkages and binding across distributed data bases over the cloud.