Research On Multi-Sensitive Attributes Data Publishing Grouping Method For Privacy Preserving

In recent years,privacy protection has always been a concern in the process of data publishing,in particular,medical data often contain a lot of sensitive information about patients,such as disease,patient doctor etc.At present,most of the researches on multiple sensitive attributes are based on L-diversity model :Connect to the quasi identification table and the sensitive information table using a lossy connection technique,or generalize and hide the identifier information,which leads to greater loss of data.In order to solve these problems,this paper proposes two grouping algorithms based on the sort of major sensitive attribute and the similar quasi identifier.The main process of SMSA: Constructing multidimensional bucket structure based on sensitive attributes,the data records is mapped to the multidimensional bucket structure according to the sensitive attribute value,then,groups are based on multidimensional buckets,select the primary sensitive attributes,calculate the dimension capacity of the main sensitive attributes,and then traverse each bucket corresponding to each of the main sensitive attributes according to the size of the dimension and meeting the L-diversity.The main process of SQI: cluster the data set,group in each data set generated by the cluster,select a data record,calculate other data records to its distance,sort according to the distance,select the small group as far as possible in one group by distance,let the group meet L-diversity.The experiment results show that loss of information,concealment and loss of additional information,three data quality indicators are low,this method can reduce the generalization and concealment of the non sensitive attributes and improve the availability of data.