Measuring the Quality of Domain Models Extracted from Textbooks with Learning Curves Analysis

Alpizar-Chacon, Isaac; Sosnovsky, Sergey; Brusilovsky, Peter

doi:10.1007/978-3-031-36272-9_75

Cited by 2 publications

(1 citation statement)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…KC-based learner models have been used in early tutoring systems in programming environments, such as the LISP tutor [3], for individualized problem selection [1]. Recently, Alpizar-Chacon et al used concepts extracted from textbooks to manually annotate code reading exercises as a possible domain model with promising results [2]. However, these KC models were time-consuming and limited in scalability as they required manual annotation of all problems by an expert.…”

Section: Kc Models For Programming Datamentioning

confidence: 99%

Reexamining Learning Curve Analysis in Programming Education: The Value of Many Small Problems

Demirtaş,

Fowler,

Cunningham

2024

Preprint

View full text Add to dashboard Cite

[to appear in EDM 2024] Analyzing which skills students develop in introductory programming education is an important question for the computer science education community. Knowledge components, units of knowledge that can be measured by performance on a set of tasks, can be used to identify key concepts and skills in a given domain. While knowledge components in other domains have been successfully identified using learning curve analysis, such attempts on students' open-ended code-writing assignments have not been very successful. To understand why, we replicated a previously proposed approach, which uses abstract syntax tree (AST) nodes as knowledge components, on data collected across multiple semesters of a large-scale introductory programming course. Findings from our replication show that, given sufficient measurement opportunities, AST nodes can provide a viable knowledge component model for learning curve analysis to understand student learning, contrary to earlier findings. In addition to providing evidence for the validity of AST-based knowledge components, we recommend a set of conditions for programming courses that enable knowledge components generated using AST nodes to be successfully observed using learning curve analysis. This supports the integration of learning curves as a tool in any environment collecting code-writing data to be used for modifying the curriculum based on student performance and monitoring the development of skills related to language elements.

show abstract

Section: Kc Models For Programming Datamentioning

confidence: 99%