Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Building Contextual Knowledge Graphs for Personalized Learning Recommendations using Text Mining and Semantic Graph Completion

Hasan Abu-Rasheed, Mareike Dornhöfer, Christian Weber, Gábor Kismihók, Ulrike Buchmann, Madjid Fathi

TL;DR

The paper addresses the need for context-aware personalization in learning platforms by transforming hierarchical learning-object models into contextual knowledge graphs through a custom text-mining pipeline that extracts semantic relations from bilingual LO descriptions. The authors introduce a 5-level taxonomy and extend it with semantic edges, enabling LO connections across learning scenarios to support contextual recommendations. They validate the approach with a combination of quantitative graph-quality metrics and qualitative expert evaluations, showing increased LO connectivity, more distinct contextual communities, and higher betweenness centrality compared to the original hierarchy. The work demonstrates that semantic KG completion can robustly represent LO context, though it relies on rich textual metadata and faces multilingual challenges, suggesting directions for improving robustness and domain-specific feature integration.

Abstract

Modelling learning objects (LO) within their context enables the learner to advance from a basic, remembering-level, learning objective to a higher-order one, i.e., a level with an application- and analysis objective. While hierarchical data models are commonly used in digital learning platforms, using graph-based models enables representing the context of LOs in those platforms. This leads to a foundation for personalized recommendations of learning paths. In this paper, the transformation of hierarchical data models into knowledge graph (KG) models of LOs using text mining is introduced and evaluated. We utilize custom text mining pipelines to mine semantic relations between elements of an expert-curated hierarchical model. We evaluate the KG structure and relation extraction using graph quality-control metrics and the comparison of algorithmic semantic-similarities to expert-defined ones. The results show that the relations in the KG are semantically comparable to those defined by domain experts, and that the proposed KG improves representing and linking the contexts of LOs through increasing graph communities and betweenness centrality.

Building Contextual Knowledge Graphs for Personalized Learning Recommendations using Text Mining and Semantic Graph Completion

TL;DR

The paper addresses the need for context-aware personalization in learning platforms by transforming hierarchical learning-object models into contextual knowledge graphs through a custom text-mining pipeline that extracts semantic relations from bilingual LO descriptions. The authors introduce a 5-level taxonomy and extend it with semantic edges, enabling LO connections across learning scenarios to support contextual recommendations. They validate the approach with a combination of quantitative graph-quality metrics and qualitative expert evaluations, showing increased LO connectivity, more distinct contextual communities, and higher betweenness centrality compared to the original hierarchy. The work demonstrates that semantic KG completion can robustly represent LO context, though it relies on rich textual metadata and faces multilingual challenges, suggesting directions for improving robustness and domain-specific feature integration.

Abstract

Modelling learning objects (LO) within their context enables the learner to advance from a basic, remembering-level, learning objective to a higher-order one, i.e., a level with an application- and analysis objective. While hierarchical data models are commonly used in digital learning platforms, using graph-based models enables representing the context of LOs in those platforms. This leads to a foundation for personalized recommendations of learning paths. In this paper, the transformation of hierarchical data models into knowledge graph (KG) models of LOs using text mining is introduced and evaluated. We utilize custom text mining pipelines to mine semantic relations between elements of an expert-curated hierarchical model. We evaluate the KG structure and relation extraction using graph quality-control metrics and the comparison of algorithmic semantic-similarities to expert-defined ones. The results show that the relations in the KG are semantically comparable to those defined by domain experts, and that the proposed KG improves representing and linking the contexts of LOs through increasing graph communities and betweenness centrality.
Paper Structure (15 sections, 3 figures, 1 table)

This paper contains 15 sections, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Background
  3. Contextual Knowledge Graph Construction
  4. Text Mining for Semantic Relation Extraction
  5. Knowledge Graph Construction
  6. Evaluation and Results
  7. Evaluation Dataset
  8. Evaluation of the Semantic Relation Extraction
  9. KG Evaluation Metrics
  10. Average Degree Centrality (ADC)
  11. Clustering Coefficient (CC)
  12. Weakly Connected Components (WCC)
  13. Betweenness Centrality (BC)
  14. Qualitative expert-evaluation
  15. Conclusion

Figures (3)

  • Figure 1: Text mining pipeline (TMP) and semantic similarity calculation for title and description properties of the learning objects.
  • Figure 2: First three levels of the KG structure, representing the role of semantic relations in creating the KG from the hierarchical one.
  • Figure 3: Semantic-relation textual-similarity scores (Purple), compared to the Journeys (Blue and Orange) connected by those relations through the TMP.