Sri Lanka institute of Information Technology -2020
2020-106
Index Terms: Knowledgebase, Abstract Syntax tree, natural
language generation, image processing, machine learning, antlr
Introduction Results & Discussion
This research investigates and provides insight into the applicability and
Analyze answer paper
development of java based online exam evaluator as a solution to
The tuple of each table stored in the database is mapped to the single
traditional onerous manual exam assessment methodology. It allows
ontology based on the generated ontology model. The mapping matches
students to take online exams in Java, automatically reporting the results
each tuple to the entity and asserts the corresponding Named entity to
to the administrator simultaneously. This software application is used
the table to which the tuple belongs
Knowledgebase, Abstract Syntax tree (AST), Natural Language Generation
Mark sheet
(NLG), Image processing, and Machine Learning (ML) as key technologies.
The source code of the student answer sheet, and the source code of the
Literature Survey
marking scheme will be analyzed. Using Abstract Syntax Tree concept.
Upon conducting a review on available literature, several approaches
Total marks and grade will be calculated automatically
towards developing an automatic Java programming code evaluation
Suggestion report
system can be identified. These proposed systems include plug-in code
defect detection, mutation and penetration testing methods, semi- Using ANTLR with the help of the AST concept by analyzing the source
automatic or automatic test code generating tools and various Computer- codes of both the answer sheet and the marking scheme. A suggestion
aided software engineering (CASE) tools. report will be generated using the all suggestions.
Research Problem
Occupying much time, the occurrence of errors while manually checking, and
the inability to check manually the whole lot of codes effectively that come
across in a Java answer script. Also, the examiners getting bored for such a
heavy mess of work. Model prediction
A classification report will be generated from the sample model as a
result. Test data will be predicted and exported to a CSV file.
Possible Solution
Objectives
The research group suggests the ‘Online Exam Evaluator’ which is specialized
in allowing students to answer questions that are available in the system and A. Construct knowledgebase and Ontology.
automatically evaluate that answer by comparing it with the model answer. B. Marks Allocation.
Through this process, C. Generate a suggestion report.
D. View suggestion report and excel report.
it solves the issues occurred when doing it manually rather than adopting an
E. Create a correct sample model.
automated system. A special function will be developed to predict pass and fail
status of students.
Conclusion
Overview
The system has been developed with much care that it is free of errors
Diagram and at the same time, it is efficient and less time-consuming. The
important thing is that the system is robust. Also, provision is provided for
future developments in the system. The entire system is secured.
Providing automatic assistance to the evaluator will improve on the
machine to a human agreement and solves the problems of automatic
grading systems. The student can see his/her result accurately without
any influence
Methodology
• Lecturer upload question paper and Marking scheme. Using set of structural References
data construct knowledge base
[1] H. Zhu and D. Jin, “Java source code defect detection,” in 2015
• Source code of the student answer sheet analyze to compare with source 8thInternational Conference on Biomedical Engineering and
code of the marking scheme using AST concept. Generate an excel sheet Informatics(BMEI), 2015, pp. 695–699.
report to show student grade and marks [2] F. Mariya and D. Barkhas, “A comparative analysis of mutation
testingtools for Java,” in 2016 IEEE East-West Design Test
• Compare the source code of the student answer sheet and marking scheme Symposium(EWDTS), 2016, pp. 1–3.
using ANTLR. Generate suggestion report and let student to view suggestion [3] S. Jain, R. Johari, and A. Kaur, “PJCT: Penetration testing based
JAVAcode testing tool,” in International Conference on
report using student ID (using OCR concept)
Computing,Communication Automation, 2015, pp. 800–805.[4] N.
• Predict the complexity of a java online exam by creating a sample model using Funabiki, R. Kusaka, N. Ishihara, and W. Kao, “A proposal of testcode
generation tool for Java programming learning assistant system,”in 2017
machine learning algorithms. Logistic regression has been used to generate
IEEE 31st International Conference on Advanced InformationNetworking
the F1 score which helps for the prediction. and Applications (AINA), 2017, pp. 51–56.
[4] Cheng-hui Huang and Huo Yan Chen, “A semi-automatic generator
forunit testing code files based on JUnit,” in 2005 IEEE
InternationalConference on Systems, Man and Cybernetics, 2005, vol. 1,
pp. 140-145Vol. 1.[6] E. Albert, I. Cabanas, A. Flores-Montoya, M.
Gomez-Zamalloa, and S.Gutierrez, “jPET: An Automatic Test-Case
Generator for Java,” in 201118th Working Conference on Reverse
Engineering, 2011, pp. 441–442
.[5] D. Al-Ashwal, E. Zaid Al-Sewari, and A. Abdulghani Al-Shargabi,
“ACASE tool for JAVA programs logical errors detection: Static
anddynamic testing,” in 2018 International Arab Conference on
InformationTechnology (ACIT), 2018, pp. 1–6
