Entity Resolution and Information Quality
Entity Resolution and Information Quality
| Hi-index | 0.00 |
Data sets containing information for an overlapping group of real-world identities present a very high likelihood that the identifying attributes and attribute values for these identities may be inconsistent between the data sets. Differences in the types of identifying attributes or attribute values inhibit proper record linkage and identity resolution. Traditional approaches to record linkage are commonly utilized however the results from these approaches do not demonstrate the highest possible levels of confidence and utility. Syntax, semantics, and temporal aspects of data sets should be understood and incorporated into the methodology of heterogeneous data set integration. Domain-specific expertise is a key component of methodology development. The goal of this research is to determine a course of action which will facilitate knowledge-driven identity-resolved longitudinal data studies with optimal record linkage for data sets containing varying identifying attributes and attribute values obtained through various collection methods over a number of years. The proposed identity resolution methodology will be demonstrated with four years of actual education data for students within Arkansas Department of Education data sets. This research will facilitate a FERPA-compliant plan for resolving the representations of real-world identities across multiple longitudinal education data sets, allowing for record linkage of statewide education data and increasing the capability of various state agencies to coordinate future research efforts for education data.