Scholarlink

<p>In recent years, Learning Analytics has become a key discipline for enhancing educational outcomes through the collection, analysis, and interpretation of data related to learners and their learning environments. Modern educational institutions nowadays typically operate multiple e-learning platforms, such as Learning Management Systems, Student Information Systems, Massive Open Online Course platforms, and virtual classrooms, which generate extensive historical records. These records can offer valuable insights into student performance and academic trends if properly analysed. A significant challenge is that each of these e-learning datasets is maintained separately and not integrated with the others. This paper presents a mixed-method data retrieval model designed to support learning analytics and student performance monitoring in Higher Education Institutions. The proposed model, known as the Heterogeneous Student Performance (HSP) data retrieval model, integrates semantic technologies with SPARQL query generation to enable accurate and automated retrieval of student data from heterogeneous sources. Competency Questions (CQs) derived from thematic analysis of domain expert interviews are used to guide ontology development and query formulation. Each CQ is translated into SPARQL queries through structured SPO triples, enabling precise mapping of learning and assessment data. The model comprises ten structured phases, including ontology development or reuse, validation, schema mapping, and SPARQL query validation. Experimental validation demonstrates the model’s effectiveness in generating relevant data insights, supporting decision-making for curriculum review, student support, and institutional performance evaluation, with the accuracy, precision, recall, and F1 score of the model being above <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\( \ge 99.1\%\)</EquationSource> </InlineEquation>, <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\ge 98.4\%\)</EquationSource> </InlineEquation>, <InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(\ge 100\%\)</EquationSource> </InlineEquation>, and <InlineEquation ID="IEq4"> <EquationSource Format="TEX">\(\ge 99.1\%\)</EquationSource> </InlineEquation>, respectively.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Advancing semantic search in higher education through the Heterogeneous Student Performance data retrieval model

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月26日

查看全文
  • Mohd Hafizan Musa,
  • Sazilah Salam,
  • Mohd Adili Norasikin,
  • Wendy Hall,
  • Jack Febrian Rusdi

摘要

In recent years, Learning Analytics has become a key discipline for enhancing educational outcomes through the collection, analysis, and interpretation of data related to learners and their learning environments. Modern educational institutions nowadays typically operate multiple e-learning platforms, such as Learning Management Systems, Student Information Systems, Massive Open Online Course platforms, and virtual classrooms, which generate extensive historical records. These records can offer valuable insights into student performance and academic trends if properly analysed. A significant challenge is that each of these e-learning datasets is maintained separately and not integrated with the others. This paper presents a mixed-method data retrieval model designed to support learning analytics and student performance monitoring in Higher Education Institutions. The proposed model, known as the Heterogeneous Student Performance (HSP) data retrieval model, integrates semantic technologies with SPARQL query generation to enable accurate and automated retrieval of student data from heterogeneous sources. Competency Questions (CQs) derived from thematic analysis of domain expert interviews are used to guide ontology development and query formulation. Each CQ is translated into SPARQL queries through structured SPO triples, enabling precise mapping of learning and assessment data. The model comprises ten structured phases, including ontology development or reuse, validation, schema mapping, and SPARQL query validation. Experimental validation demonstrates the model’s effectiveness in generating relevant data insights, supporting decision-making for curriculum review, student support, and institutional performance evaluation, with the accuracy, precision, recall, and F1 score of the model being above \( \ge 99.1\%\) , \(\ge 98.4\%\) , \(\ge 100\%\) , and \(\ge 99.1\%\) , respectively.