The Research Of Privacy Protection Methodologies On The Trusted Database

With the development of the society and the emphasis on the privacy, the research of the privacy database has been one of the hotspots of database security. The privacy database provides some protection mechanism for privacy data usage based on the privacy preferences defined by the privacy data providers. Compared with satefy database, the privacy database should protect the privacy data as well as be easy to use them. The reason why the privacy data protection lacks of uniform protection formula is that the protection for privacy data relies on the preference of privacy data owner. This dissertation is about studies on protection methodologies of multi-owner privacy data and implementation, which includes: the concept and taxonomy of multi-owner privacy data; the methodology of modeling multi-owner privacy data by object deputy model; the privacy data protection language in privacy database based on TOTEM; multi-owner privacy data protection policy detection mechanism and etc. Research contents and innovations of this dissertation are summarized as follows:1. The privacy concept and formal model for Hippocratci databaseNowadays, the research on privacy database focuses on that have single centralized owners. In fact, another kind of privacy data, which have multiple owners, extensively exists in our daily life. This kind of data usually comes into being in the interactions of multiple individuals, and they will expand in this procedure. Obviously, this kind of data should be treated much carefully. In my thesis, this kind of data and its hiberarchy is described, and the privacy protection demands are reviewed.2. Methodology of modeling privacy by object deputy mechanismPrivacy data of multi-owner includes the protection demands of the single owner data type, including polymorphism and individuation. It also asks for more demands in data evpOep enWassRciaWe reguOaWRn's prrwcWON, and correspondence of policies defined by multiple defference owners. The single owner privacy daWp Rdeong is Eased Rn We "view" mechanism in the relationship databases. This is a very straightforward method, but cannot meet the demands of multi-owner privacy data type. In the thesis, a multi-owner privacy data modeling method based on the object deputy mechanism is brought forward, which has the merits of relational data models as well as that of object-oriented data models. In the multi-owner privacy data model, the basic objects are the initial privacy data, deputy objects are the different data versions designated by different data owners in different ratings. In this kind of data model, the new data, which are produced in semantic expanding, are defined as deputy data, too. This method can agilely define the attributes of any objects based on the deputy mechanism, generate the new data types based on the relationship of individual basic objects, as well as maintain the coherence of different data versions by objects renovating transference. For the sake of deputy class explosion, we define the switching manipulation to extend the deputy mechanism. In my thesis, the implementation of these methods is described.3. The privacy data protection language: design and implementationIn the research of privacy database, there is a basic hypothesis that data provider can designate the particular privacy demand for every privacy data type. In the fact, because of the complicated semantic scene, this hypothesis may not be satisfied. In the thesis, an architectonic of privacy data protection language is designed, referring with the EPAL regulations and SQL standards. A demo of the privacy data protection language in TOTEM database is shown in the thesis.4. Multi-owner privacy policy conflict detection mechanismIn the single owner privacy protection, the data creator and data owner is same individual. However, for the multi-owner, the creator and the owner may be different. Further more, in the procedure of data evolution; there will be new owners, and new demands. The privacy protection mechanism should meet the demands of every individual and that of the all individuals. In the thesis, a method of privacy data protection policy detection mechanism based on the sub-graph isomorphic is provided, which is to detect the collision among different owners. The bi-directional pointer in object deputy model can link all privacy policies defined for the privacy data and its different data versions and evolutional versions by defference owners. These policies form a policy set. Each policy in the set is abstracted as a directed edge, and the policy can be abstracted as a stratified-directed graph. Each policy constraint, which is defined by an owner and used to represent the prohibited privacy data release pattern of the owner, is abstracted as a stratified-directed subgraph. The method how to analyze and model the policy constraints is discussed in the section and an algorithm is proposed to detect whether the stratified-directed subgraph of a privacy constraint mode is isomorphic to the stratified-directed graph of a privacy policy set.5. The design and implementation of privacy data protection module in the trust databaseIn the TOTEM, an object deputy database system, there are privacy data protection modules. The methods discussed in my thesis, will be used in the trust database system in TOTEM circumstance. Discussed privacy data protection mechanism will be a part of protection module for emails. In the experimentation, the feasibility and validity of privacy data protection of multi-owner is proved the truth.