Linked data technologies to support higher education challenges: student retention, progression and completion

Abstract

Around the world, higher education institutions are facing a growing number of challenges. In recent decades, considerable interest has emerged on identifying those challenges and proposing efficient ways to address them. This thesis reviews a wide range of literature on higher education challenges and identifies related intuitional data, data repositories and external open data sources to address these challenges. It subsequently explores whether certain higher education challenges and in particular student retention, progression and completion can be better addressed using data from various data sources and the recent development of technologies such as, data analytics and linked data. Traditionally, research in this area is survey-based and survey-based studies have some drawbacks such as, low participation rate and the high cost associated with it. This research sought to overcome these problems. To this end, two experiments were conducted. The first experiment examined the sufficiency of linked data and external open data sources to develop blended prediction models to predict at-risk students in their first year of study. The result based on 149 undergraduate students’ data, established that prediction models based on institutional repositories and external open data perform better than survey-based one. The second experiment examined the capabilities of institutional repositories and external open data sources in predicting students’ first year marks and established that models using institutional repositories and external open data sources can perform better than models based on only institutional repositories. In order to examine the capabilities of linked data, external open data and data analytics, a data integration and analytics environment was deployed. The four key contributions of this thesis are: (1) it presents a comprehensive list of higher education challenges and required data and data repositories to address these challenges; (2) it demonstrates how external open data sources can be used to accurately predict students at-risk and students’ first year marks; (3) it shows how including external open data sources in prediction models can increase the overall model accuracy and (4) it establishes the strengths and weaknesses of linked data to support in employing data analytics for predictive models in student retention, progression and completion.

More information

Identifiers

ORCID for Thanassis Tiropanis: orcid.org/0000-0002-6195-2852

Catalogue record

Export record