Skip to main content Skip to navigation

Task Synthesis for Elementary Visual Programming in XLogoOnline Environment

Project Overview

The document explores the integration of generative AI in education through the development of XL OGO SYN, a technique that synthesizes high-quality programming tasks for the XLogoOnline platform, which combines visual programming with the Logo language for novice learners. By generating tasks with varying difficulty levels, XL OGO SYN enhances the educational experience by offering personalized practice opportunities that cater to individual learner needs. The initial implementation of these synthesized tasks has shown promising results, significantly improving learners' success rates in grasping programming concepts. Overall, the use of generative AI in this context demonstrates its potential to facilitate tailored learning experiences and foster greater understanding in programming education.

Key Applications

XL OGO SYN

Context: XLogoOnline platform for novice learners in programming education

Implementation: Utilizes symbolic execution and constraint satisfaction techniques to generate tasks at varying difficulty levels based on a reference task.

Outcomes: Improved learner success rates on subsequent reference tasks, with initial data showing a higher success rate for learners who engaged with synthesized tasks.

Challenges: Difficulty in generating more complex tasks; current difficulty categorization may not align with learners' perceptions.

Implementation Barriers

Technical Barrier

Generating high-quality tasks is time-consuming and requires synthesizing from a large pool of tasks.

Proposed Solutions: Future work may explore learning-based strategies and generative AI models to accelerate the synthesis process.

Personalization Barrier

XL OGO SYN does not incorporate individual learner's code, limiting task personalization. Future developments could focus on personalizing tasks based on learners' misconceptions.

Proposed Solutions: Enhancing task personalization through integration of individual learner data and addressing misconceptions.

Difficulty Perception Barrier

The predefined rules for task difficulty may not match learners' understanding of what constitutes difficulty. This necessitates refining the notion of task difficulty through analysis of learners' interactions with the platform.

Proposed Solutions: Refining the notion of task difficulty based on learner interactions and feedback.

Project Team

Chao Wen

Researcher

Ahana Ghosh

Researcher

Jacqueline Staub

Researcher

Adish Singla

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Chao Wen, Ahana Ghosh, Jacqueline Staub, Adish Singla

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies