Rika Yoshii, Ph.D. and Jacquelyn Hernandez [email protected]CSIS Department California State University, San Marcos Send us suggestions and requests to use the system ItsLEADR: INTELLIGENT TUTORING SYSTEM FOR LEARNING ENGLISH ARTICLES BY DIAGRAMMATIC REASONING CALICO 2015 send comments to [email protected]
31
Embed
Rika Yoshii, Ph.D. and Jacquelyn Hernandez [email protected] CSIS Department California State University, San Marcos Send us suggestions and requests to.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Rika Yoshii, Ph.D. and Jacquelyn Hernandez [email protected] DepartmentCalifornia State University, San Marcos
Send us suggestions and requests to use the system
ItsLEADR: INTELLIGENT TUTORING SYSTEM FOR LEARNING ENGLISH
•IS-link is used when a noun phrase refers to the noun node itself.•IS-IN-link is used when a noun phrase refers to some member(s) of the noun node.•IS-ALL-OF-link is used when a noun phrase refers to all members of a noun node.•IS-REP-link is used when a noun phrase refers to a representative of a noun node.
3 phases: Introduction – introduces intent types, diagrams and their components and asks review questions to make sure the student understands them.
Diagram selection - an intent type and a sentence are given, and the student is asked to choose the correct diagram.
Article selection - an intent type and a diagram are given, and the student is asked to fill in a blank space in the exercise sentence with an article.
If the Help button is clicked, then a pop up window reviewing the diagrams and their components will appear.
PEDAGOGY
The student cannot move on to the next intent type until the mastery for that type is demonstrated by repeatedly answering correctly.
The student cannot move onto the next phase until the mastery of that phase is demonstrated..
perturbedversion of the domain model.Predictability model – probabilistic inference for predicting the student answers based on performance.
Hint table - composed of the question type, the correct answer, the expected student answer, and corresponding hints addressing the student’s misconception.
STUDENT MODELING
If the expected student answer is an incorrect answer, the system will retrieve a hint from the hint table to present along with the question to the student.
If the student answers incorrectly, the system will retrieve a hint from the hint table to help the student.
Performance record is always updated.
Once the student has successfully mastered a phase, the system computes a student performance summary.
Average of 3.44 (5 being the best) when rating their belief that the ItsLEADR system is useful.
Commented that the diagrams help to give context where as memorization was the main mechanism in classrooms.
Found the links to be diffi cult to understand and remember.
Most participants did not read the introductory information and relied on the Help button.
EVALUATION RESULTS
Revise the links and/or link definitions and images to make them easy to understand and remember.
Create a training manual written in English.
Add online help features such as balloons and visual walkthrough.
SUGGESTED BY EVALUATORS
Enhanced DaRT from a CALL system to an intelligent tutoring system.
Refined the intent types.Improved the diagrams.Individualized tutoring by predicting
student answers.Developed in C++ with Qt for GUI.Performed preliminary evaluation.
SUMMARY
Enhance the prediction of student answers by determining the root cause of errors and use this information in giving hints and selecting remedial exercises.
A larger scale formative evaluation followed by further improvements.
A summative evaluation with experimental groups and a control group, comparing improvements and retention over several months.
Please send suggestions and requests to use the system to [email protected]