Privacy-Preserving Temporal Record Linkage
Record Linkage (RL) is the process of linking records representing the same real-world entity in one or more databases. Since no unique identifier of the entities to be linked is available and shared across databases, Record Linkage requires sophisticated comparisons between a set of attributes, called quasi-identifiers (QIDs), to compute the similarity between pairs of records. Such QIDs often contain private and confidential entity information, and therefore revealing or exchanging them to perform Record Linkage is not possible for privacy and confidentiality reasons. Privacy-Preserving Record Linkage (PPRL) aims to address this problem by identifying and linking records that correspond to the same real-world entity in different databases without revealing any sensitive information about these entities. The idea behind PPRL techniques is to perform, at the data source level, a process of pseudonymization, by which personal identifiers are masked (encoded) to make people unidentifiable; Record Linkage will then be performed on such pseudonymized data. In one of the most popular protocols for PPRL between two databases, known as the Three-Party Protocol, a (trusted) third party performs Record Linkage on the pseudonymized data received from the database owners.
In both traditional Record Linkage and PPRL, the databases to be linked are considered static, that is, the records do not change over time. However, entities may change, and their attribute values evolve over time. Incorporating temporal information into similarity calculations between pairs of records can help identify similar records that belong to the same entity over a period of time. Temporal Record Linkage (TRL) computes similarities between pairs of records based on temporal information. The aims of Privacy-Preserving Temporal Record Linkage (PPTRL) is to use of sensitive information in TRL while ensuring the privacy of that information.
Categorie: DBGroup Activities