Smart Grading Learning Management System for Software Engineering and System Programming Courses

Project Domain / Category:

Web Application

Abstract/Introduction

The transition to digital education has introduced new possibilities and challenges, especially in the formative assessment of students. Manually grading assessments, particularly open-ended responses, is both time-consuming and prone to human bias. Evaluating subjective and System Programming-based questions requires substantial effort from educators, often resulting in delayed feedback, inconsistencies, and limited personalized attention for each student.

To overcome these challenges, we proposed Smart Grading Learning Management System (LMS) for Software Engineering and System Programming Courses. The system uses advanced Natural Language Processing (NLP) techniques along with Machine Learning (ML) and Deep Learning (DL) models for evaluating student deliverables. Through smart grading and real-time feedback, the system enhances student engagement in Software Engineering and System Programming courses.

Functional Requirements:

The Smart Grading Learning Management System for Software Engineering and System Programming Courses evaluate both objective and subjective questions. The system uses consistent test administration and scheduling throughout the semester. The following are the functional requirements.

The proposed system will have the following main users:

Admin Teacher and Student.

1.      Registration module: It will facilitate the registration process for students and teachers. Admin will approve and perform activation of the students and teachers accounts and registration requests.

2.      Login Module: After successful registrations, all types of users will be able to login to the system by using their registered email and password.

3.      Your application will assist the teacher with the auto scheduling of quizzes and assignments of different subjects including Software Engineering and System Programming courses.

4.      The system can generate objective and subjective questions for Software Engineering course, including sub domain: Software Process Modeling and Software Architecture and Design.

5.      System can generate objective and subjective questions for System Programming course, including sub domain: Low-Level Programming & Assembly Language and Networking & Socket Programming.

6.      Teachers can compile a labeled dataset of existing objective and subjective questions along with their correct answers for both domains including Software Engineering and System Programming.

7.      The Learning Management System can automatically generate objective and matching questions while also providing an option to define the difficulty level and cognitive level complexity for each subjective question.

8.      The Learning Management System can utilize the Mohler dataset for analytical purposes.

9.      Learning Management System can align grading with pre-defined rubric to ensure consistency.

10. In Software Engineering course, the system can leverage transformer-based Deep Learning models like BERT, GPT, RoBERTa and DistilBERT and XLNet to assess the similarity between a student's answer and a model/desired answer.

11. In the Software Engineering course Learning Management System can assign scores based on semantic similarity by understanding context, meaning and semantic of subjective questions rather than simple keyword matching.

12. For System Programming course, the system should use OpenAI Codex and PolyCoder to automatically evaluate students' code by compiling, analyzing and validating solutions for programming tasks.

13. The system should evaluate System Programming assignments by compiling and executing code and verify correctness, efficiency, and relevance to system programming concepts.

14. The system can deliver personalized feedback to students by highlighting their mistakes and providing guidance on areas for improvement.

15. The System should allow teachers to export grades and performance reports in common formats (e.g., CSV, Excel, PDF).

16. The system should evaluate the accuracy of various Deep Learning and Machine Learning models by using performance metrics such as Mean Squared Error (MSE) and Root Mean Squared Error (RMSE).

17. The System firstly collects each set of student answers then performs automatic evaluation and compares the predicted grades (generated by the model) with the actual grades assigned by instructors. It then applies Mean Squared Error (MSE) and Root Mean Squared Error (RMSE) to assess the model's accuracy.

18. The system should enable teachers to set MSE and RMSE threshold values limits. If these limits are exceeded, it should trigger an alert for potential manual review.

19. The system should offer teachers metrics for each question, such as the average score, standard deviation, and difficulty indices (the percentage of students who answered correctly).

20. The system should automatically identify and flag questions with a high incorrect response rate, suggesting that questions may be challenging or unclear to most students.

21. Teachers should be able to use classification tools on both domains Software Engineering and System Programming and should categorize questions after assessments by labeling them as "easy”, “medium" or "hard".

22. Teachers can generate reports showing how students performed on each question.

23. Teachers be able to create rubrics for different types of subjective questions and programming questions (e.g., essays, short answers, case studies and their implementations).

24. Students can attempt the assignments allocated for the given course by the teacher.

25. Ensure the platform is optimized for mobile use, allowing students to access learning materials and assessments on a variety of devices, including smartphones and tablets.

26. Provide a help center, FAQs, or chat support to assist students with any issues they encounter or questions they may have about the platform.

Tools: JSP, PHP, Python, JavaScript/HTML/CSS, MySQL, PyTorch, OpenAI Codex, PolyCoder, Keras, Padas, TensorFlow, BERT, RoBERTa, DistilBERT, XLNet, Transformer.

Supervisor:

Name: Muhammad Umar Farooq

Email ID: umarfarooq@vu.edu.pk

Skype ID: live:umarvc