Bolstering independent learning in the online age: concept-centric approaches to improving content accessibility

As learning becomes more and more dependent on online Web sources and online learning platforms as evidenced by the shift to remote learning during COVID-19, new challenges are emerging in the educational landscape. In this thesis, we will explore several new challenges in the shifting landscape of remote learning, specifically as they apply to independent and ad-hoc learners. Among these challenges are: 1) ease of access to material, 2) support for browsing online educational material, and 3) scaffolding for supporting the needs of learners in the form of proactive anticipation of confusion and problem areas, and finally 4) implementing our proposed solutions for end users online. In order for our work to have a wider impact and general applicability, we take a ``concept centric" approach to the above mentioned challenges. Concepts are the basic building blocks for any learning experience: 1) learning modules can be divided into a series of concepts discussed in the lectures, 2) from a browsing perspective, learners enter concept phrases as search queries when exploring new material, 3) student comprehension is measured by how well they understand the concepts covered, and 4) assessment are designed such that each item or question corresponds to a concept taught in the learning module. Thus taking a concept-centric approach enables us to explore a wide variety of problems facing independent learners. It also enables our work to be easily interpretable to non-domain experts and users of all backgrounds. Following the concept-centric approach, this thesis makes the following contributions, synergistically advancing technologies for supporting independent online learners: 1) We propose and study a new approach for extracting concepts from online educational materials using back-of-the-book indexes as training data, 2) We propose and study a new algorithm to automatically construct a Browsable Concept Index using lecture transcript data from online courses, 3) We study how to effectively scaffold independent learning by Modeling Difficult Concepts using clickstream data from lecture watching sessions in order to anticipate student needs in problem areas in online courses, and finally, 4) We develop a novel system (called SmartMOOCs) to enable learners to use the proposed algorithms and evaluate it via a user study. The user study demonstrated that the Browsable Concept Index (BCI) in SmartMOOCs improved participants' ability to locate and engage with course concepts. Higher satisfaction and better performance in both navigation efficiency and post-task concept recall were observed among participants, which supported SmartMOOCs' effectiveness in fostering targeted, independent learning.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois Urbana-Champaign

Year dc:date

2025

Author dc:creator

• Boughoula, Assma

Contributors dc:contributor

• Zhai, ChengXiang
• Zilles, Craig
• Lewis, Colleen
• Alawini, Abdussalam
• Zapata-Rivera, Diego

• concept extraction
• concept graph
• concept map
• learning support
• learning scaffolding
• independent learning
• smartmoocs

Statement dc:rights

• Copyright 2025 Assma Boughoula

Language dc:language

en

Handle dc:identifier

https://hdl.handle.net/2142/132449

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/132449