Year: 2024
Type: dissertation
Abstract: Decision making is one of the most challenging processes that higher education institutions continuously experience worldwide. Most educational decisions rely mainly on evaluating the academic profile of staff members, which usually includes the academic and research activities of the teacher. The massive amount of scattered educational data, if represented in traditional forms, causes the problem of ambiguity and inaccuracy of decisions. Educational institutions have recently been attempting to apply emerging technologies in the data engineering field to solve as many challenges as possible. In addition, online libraries continuously produce an enormous amount of open scholarly data, including publications, citations, and other research activity records, which could effectively improve the quality of academic decisions when linked with the local data of universities. This thesis presents the academic profiles and course records semantically, and employs them with a scientific knowledge graph as linked data to enrich the internal data and support the decision-making process within universities. The proposed approach is applied to assign courses to the most qualified academic staff as a proof-of-concept experiment. Traditionally, this process is performed manually by heads of departments and is considered time-consuming, especially when the data are in textual format. This research aims to address this challenge. To this end, courses and academic profiles are represented semantically in RDF format, in order to improve the quality of the institutional data. To ensure the efficiency of this process, a survey is conducted to identify the key factors that influence decision making during the distribution of courses among staff members, which was successfully distributed to the heads of departments who actively participated and provided their variable insights into this matter. The survey results indicated that the research areas of academic staff and whether they had taught the course before are the most important factors that are usually considered in this type of decision. Furthermore, this study proves the importance of generating links between local data and external repositories with updated research records to improve the course–teacher assignment process. Linked data technology is applied to combine all the possible information affecting the course–teacher assignment decision from different resources, and the sufficiency of the linked data and the selection of external data are examined using data mining techniques. Two prediction models are developed to predict the most qualified academic teacher to teach each course, with the results being associated with 314 academic teachers and 119 courses from the Faculty of Computing and Information Technology at King Abdulaziz University. According to the obtained accuracy of the models, it is suggested that the performance is improved when the data are enriched with external scholarly open data using LD, with the accuracy increasing from 80.95% to 93.26% after applying LD techniques. Additionally, adding research records of the academic member improved the sensitivity of the models to 89.11% and 97.76%. These improvements demonstrate the importance of considering the research activities of academic members when distributing courses, especially when extracted from external repositories using LD.
Author: Ashour, Ghadeer Adnan D
Institution: Edinburgh Napier University
Language: English
Topic: Semantic Web and Ontologies
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: green
OpenAlex ID: W6925533145
Cites:
Cited by:
Related to: