Learner question's correctness assessment and a guided correction method: enhancing the user experience in an...

Area of Science:

• Educational Technology
• Natural Language Processing
• Computer Science

Background:

• Effective question formulation is critical for online learning systems (OLS) to recommend appropriate materials.
• Incorrectly formed questions can lead to poor recommendations, negatively impacting learning quality and user satisfaction.

Purpose of the Study:

• To propose a novel method for assessing the syntactic and semantic correctness of learner questions in an OLS.
• To improve the performance of OLS recommendations by accurately evaluating learner queries.
• To develop a framework for guiding learners toward correctly formulated questions.

Main Methods:

• A tri-gram language model was developed and trained on a corpus of 2,533 Java questions.
• The model was tested on a separate corpus of 634 Java questions sourced from diverse educational materials.
• A soft cosine similarity approach was employed to recommend corrected questions.

Main Results:

• The proposed method achieved 92% accuracy in distinguishing between correct and incorrect learner questions.
• The additional framework for guiding learners to correct their questions demonstrated 85% accuracy on real-time data.

Conclusions:

• Accurate assessment of learner question correctness significantly enhances OLS recommendation performance.
• The developed framework effectively assists learners in formulating accurate and relevant questions, improving their learning experience.