Date of Award

3-26-2015

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Information Science

First Advisor

Russel Bruhn

Abstract

This dissertation proposes an entity resolution approach that is context-sensitive, meaning it relies less on high-risk information, such as social security numbers, to discern whether records from a data set belong to the same real-world individual or to different real-world individuals. The research follows an iterative process of assessing the quality, or fitness of use, of identity data housed in an educational institution and then processing the data with the proposed context-sensitive entity resolution (ER) rule set. The efficacy of this process is demonstrated through calculation of the four-year adjusted cohort graduation rate, a prevalent longitudinal data analysis challenge for educational institutions. Research is conducted through ER activities, including data quality assessment and iterative rule development, to demonstrate development and measurement steps to accomplish four primary objectives: 1) evaluation on the quality of entity identity attributes through data quality assessment, 2) development of a context-sensitive ER approach using low-risk entity identity reference data, 3) evaluation of the context-sensitive approach through iterative rule development and analysis, and 4) application to calculate the four-year adjusted cohort graduation rate in educational institutions.

Download

Share

COinS