Study On Privacy Preserving Approaches Based On Temporal And Spatial Characteristics

With the further development of information technology, strong demands of finding useful knowledge from the massive data bring about the emergence of data analysis tools, such that there appears the concept of data mining. Nowadays, data mining and knowledge discovery technologies have made great progress. With the help of these technologies, many useful hiding knowledge or datum could be fetched out. However, there also have increasing risks when the data is open to public, as result, the privacy is fallen under threaten. Privacy-Preserving Data Mining (PPDM) has emerged to address these issues. The research of PPDM is aimed at bridging the gap between collaborative data mining and data confidentiality.In the research of PPDM, as two important basic characteristics of privacy information, temporal and spatiality are often ignored. These two elements could better help us study the actual database if been taken into account. Research of this thesis is based on this main idea.First of all, some basic concepts of data mining are brought forward. Then, we introduce and analyze some typical privacy preserving algorithms from data distribution, data modification, data mining algorithm, hiding objects and privacy preserving technology dimensions.Secondly, novel about privacy preserving data mining with the temporal and spatial constraint is proposed. The key concept, which is called data confidential level, is defined in the thesis. Privacy of data or transactions is preserved by using concept generalization hierarchy according to its data confidential level. Integrated with time and space, the data confidential level is extended to have properties of temporal and spatiality.Then, based on the former content, method is proposed integrated with temporal and spatial constraint. Further more, the method is applied to privacy preserving association rules mining. After the algorithm construction, some evaluate measures are used to testify and analyze the algorithm performance, which include information loss and execution time and algorithm utility and so on.