PLAGIARISM DETECTION IN TEXT-BASED ASSIGNMENTS USING

NATURAL LANGUAGE PROCESSING TECHNIQUE

PROJECT PROPOSAL

By

ISOLA OLUWATOBI KAOSARA

H/CS/23/1068

DEPARTMENT

COMPUTER SCIENCE HND 2

Supervisor

Dr. Mrs. Soyemi

FEBRUARY, 2025
Table of Content

1.​ Introduction Of The Proposed Project

2.​ Aim Of Proposed Project

3.​ Objective Of The Proposed Project

4.​ Literature Review

5.​ Research Methodology

6.​ Proposed Model And Tools

7.​ Contribution To Knowledge

8.​ Conclusion

9.​ References
 1.​ INTRODUCTION OF THE PROPOSED PROJECT

Plagiarism, the act of using someone else's work without proper acknowledgment, has
become a significant concern in academic and professional settings. With the increasing
availability of digital content, the ease of copying and pasting text has exacerbated the
problem. Traditional plagiarism detection tools often rely on simple string-matching
techniques, which are limited in detecting sophisticated forms of plagiarism, such as
paraphrasing or idea theft.

This project proposes the development of an advanced plagiarism detection system using
Natural Language Processing (NLP) techniques. By leveraging NLP, the system will be
capable of understanding the context, semantics, and structure of text, enabling it to identify
both direct and indirect forms of plagiarism more effectively. The proposed system aims to
provide a robust solution for educators, researchers, and institutions to maintain academic
integrity.

2.​ AIM OF PROPOSED PROJECT

The primary aim of this project is to design and implement a plagiarism detection system that
utilizes NLP techniques to identify and flag instances of plagiarism in text-based
assignments. The system will focus on detecting not only verbatim copying but also
paraphrased content, ensuring a comprehensive approach to maintaining academic honesty.

3.​ OBJECTIVE OF THE PROPOSED PROJECT

The objectives of the proposed project are as follows:

1.​ To develop an NLP-based model capable of analyzing and comparing text for
similarities.
2.​ To create a user-friendly interface for educators and students to check assignments for
plagiarism.
3.​ To evaluate the accuracy and efficiency of the proposed system using real-world
datasets.
4.​ To contribute to the field of NLP by exploring innovative methods for plagiarism
detection.
4. LITERATURE REVIEW
4.1 Traditional Plagiarism Detection Methods
Traditional plagiarism detection systems primarily rely on string-matching algorithms and
fingerprinting techniques. These methods compare text documents by identifying exact or
near-exact matches of substrings. Examples of such systems include Turnitin and Copyscape,
which are widely used in academic and professional settings.
●​ String Matching: This technique involves comparing sequences of characters
between documents. While effective for detecting verbatim copying, it fails to
identify paraphrased or semantically similar content (Hoad & Zobel, 2023).
●​ Fingerprinting: This method creates a unique "fingerprint" for each document by
hashing specific text segments. Documents with similar fingerprints are flagged as
potential plagiarism. However, this approach is limited in detecting sophisticated
forms of plagiarism, such as idea theft or heavily paraphrased text (Brin et al., 2020).

4.2 Advancements in NLP for Plagiarism Detection

Recent advancements in NLP have opened new avenues for improving plagiarism detection
systems. Techniques such as word embeddings, semantic analysis, and transformer-based
models have shown promise in understanding the context and meaning of text, enabling the
detection of more sophisticated forms of plagiarism.
●​ Word Embeddings: Word embeddings, such as Word2Vec (Mikolov et al., 2023) and
GloVe (Pennington et al., 2014), represent words as vectors in a high-dimensional
space, capturing semantic relationships between words. These embeddings can be
used to measure the similarity between texts, even when the wording differs.
●​ Semantic Analysis: Techniques like Latent Semantic Analysis (LSA) and Latent
Dirichlet Allocation (LDA) analyze the underlying meaning of text by identifying
topics and themes. These methods can detect plagiarism in cases where the text has
been rephrased but retains the same meaning (Landauer et al., 2023).

4.3 Existing Plagiarism Detection Tools

Several plagiarism detection tools are currently available, each with its own strengths and
weaknesses.
 ●​ Turnitin: A widely used tool in academic institutions, Turnitin employs
fingerprinting and string-matching techniques to detect plagiarism. While effective for
detecting direct copying, it struggles with paraphrased content (Heather, 2020).
●​ Grammarly: Known primarily as a grammar-checking tool, Grammarly also includes
a plagiarism detection feature. It uses a combination of string matching and semantic
analysis but is limited in its ability to detect complex forms of plagiarism.
●​ Copyscape: This tool is popular for detecting online content duplication. It relies
heavily on string matching and is less effective for detecting plagiarism in academic
texts.

5. RESEARCH METHODOLOGY

5.1 Requirement Analysis

The first phase involves understanding the needs of the end-users and defining the functional
and non-functional requirements of the system.

5.1.1 Functional Requirements

1.​ Text Input: The system should allow users to upload text-based assignments in
various formats (e.g., .txt, .docx, .pdf).
2.​ Plagiarism Detection: The system should detect both verbatim copying and
paraphrased content using NLP techniques.
3.​ Similarity Scoring: The system should generate a similarity score indicating the
likelihood of plagiarism.
4.​ Report Generation: The system should provide a detailed report highlighting
plagiarized sections and their sources.
5.​ User Authentication: The system should include a login mechanism for educators
and students to access their accounts securely.

5.1.2 Non-Functional Requirements

1.​ Accuracy: The system should achieve high accuracy in detecting plagiarism,
especially paraphrased content.
2.​ Performance: The system should process and analyze documents within a reasonable
time frame.
 3.​ Scalability: The system should handle a large number of users and documents
simultaneously.
4.​ Usability: The system should have an intuitive and user-friendly interface.

5.2. System Design

The software design phase focuses on creating the architecture and design of the system. This
includes defining the system's modules, data flow, and algorithms.
1.​ System Architecture:
○​ The system will follow a modular architecture with the following components:
■​ Text Preprocessing Module: Cleans and prepares the text for analysis.
■​ Feature Extraction Module: Uses NLP techniques to extract
meaningful features from the text.
■​ Similarity Detection Module: Compares texts and identifies
similarities.
■​ User Interface Module: Provides an interface for users to interact with
the system.
2.​ Data Flow Diagram:
○​ The input text is preprocessed and passed to the feature extraction module.
○​ The extracted features are compared with the database using the similarity
detection module.
○​ The results are displayed to the user through the interface.
3.​ Algorithms:
○​ Preprocessing: Tokenization, stemming, and lemmatization.
○​ Feature Extraction: Word embeddings (Word2Vec, GloVe) or
transformer-based models (BERT).
○​ Similarity Detection: Cosine similarity or Jaccard similarity for comparing
text vectors.
5.3. System Development
The development phase involves implementing the system based on the design specifications.
1.​ Text Preprocessing Module:
○​ Implement tokenization, stemming, and lemmatization using NLTK or SpaCy.
○​ Remove stop words and punctuation.
2.​ Feature Extraction Module:
○​ Use pre-trained word embeddings (e.g., Word2Vec)
○​ Convert text into numerical vectors for comparison.
3.​ Similarity Detection Module:
○​ Implement algorithms to calculate similarity scores (e.g., cosine similarity).
○​ Compare the input text with documents in the database and identify matches.
4.​ User Interface Module:
○​ Develop a web-based interface using Flask.
○​ Allow users to upload documents, view similarity scores, and see highlighted
plagiarized sections.
5.​ Database Integration:
○​ Store and retrieve documents using SQLite.
○​ Ensure efficient querying for large datasets.

5.4. System Testing and Evaluation

The system will be tested to ensure it meets the functional and non-functional requirements.
1.​ Unit Testing:
○​ Test individual modules (e.g., preprocessing, feature extraction) for
correctness.
2.​ Integration Testing:
○​ Test the interaction between modules to ensure seamless data flow.
3.​ Performance Testing:
○​ Evaluate the system's accuracy, efficiency, and scalability using real-world
datasets.
○​ Compare the system's performance with existing tools like Turnitin or
Grammarly.
 4.​ Evaluation Metrics:
○​ Use precision, recall, and F1-score to measure the system's effectiveness in
detecting plagiarism.

5. Flowchart of the Development Process

Figure 5.1: Flowchart

6. PROPOSED MODEL AND TOOLS
The proposed model will consist of the following components:
1.​ Text Preprocessing Module: For cleaning and preparing text data.
2.​ Similarity Detection Module: To compare texts and identify similarities.
3.​ User Interface: A web-based platform for users to upload and check assignments.
Tools and technologies to be used include:
●​ Python programming language
●​ NLP libraries such as NLTK, SpaCy, and Hugging Face Transformers
●​ Machine learning frameworks like TensorFlow or PyTorch
●​ Database systems for storing and retrieving text data

7. CONTRIBUTION TO KNOWLEDGE

This project will contribute to the field of NLP and plagiarism detection in the following
ways:
1.​ By developing a system that detects both direct and paraphrased plagiarism, it
addresses a significant gap in existing tools.
2.​ The proposed system will provide a practical solution for academic institutions to
combat plagiarism effectively.
3.​ The research findings will be documented and shared with the academic community,
fostering further advancements in the field.

8. CONCLUSION

The proposed project aims to revolutionize plagiarism detection by leveraging advanced NLP
techniques. By focusing on semantic understanding and context, the system will provide a
more accurate and comprehensive solution compared to traditional methods. The successful
implementation of this project will not only enhance academic integrity but also contribute to
the growing body of knowledge in NLP and machine learning.
References

1.​ Brin, S., Davis, J., & Garcia-Molina, H. (2023). Copy detection mechanisms for
digital documents. ACM SIGMOD Record, 24(2), 398-409.
2.​ Brown, T. B., et al. (2020). Language models are few-shot learners. Advances in
Neural Information Processing Systems, 33, 1877-1901.
3.​ Devlin, J., Chang, M. W., Lee, K., & Toutanova, K. (2020). BERT: Pre-training of
deep bidirectional transformers for language understanding. arXiv preprint
arXiv:1810.04805.
4.​ Hoad, T. C., & Zobel, J. (2023). Methods for identifying versioned and plagiarized
documents. Journal of the American Society for Information Science and Technology,
54(3), 203-215.
5.​ Heather, J. (2010). Turnitin.com and the scriptural enterprise of plagiarism detection.
Computers and Composition, 27(1), 15-28.
6.​ Landauer, T. K., Foltz, P. W., & Laham, D. (2023). An introduction to latent semantic
analysis. Discourse Processes, 25(2-3), 259-284.
7.​ Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2020). Efficient estimation of word
representations in vector space. arXiv preprint arXiv:1301.3781.
8.​ Pennington, J., Socher, R., & Manning, C. D. (2020). GloVe: Global vectors for word
representation. Proceedings of the 2014 Conference on Empirical Methods in Natural
Language Processing (EMNLP), 1532-1543.
References

●​ Alzahrani, S. M., Salim, N., & Abraham, A. (2012). Understanding plagiarism

linguistic patterns, textual features, and detection methods. IEEE Transactions on
Systems, Man, and Cybernetics, Part C (Applications and Reviews), 42(2), 133-149.
●​ Brin, S., Davis, J., & Garcia-Molina, H. (1995). Copy detection mechanisms for
digital documents. ACM SIGMOD Record, 24(2), 398-409.
●​ Brown, T. B., et al. (2020). Language models are few-shot learners. Advances in
Neural Information Processing Systems, 33, 1877-1901.
●​ Clough, P., Gaizauskas, R., & Piao, S. S. (2002). Measuring text reuse. Proceedings of
the 40th Annual Meeting on Association for Computational Linguistics, 152-159.
●​ Devlin, J., Chang, M. W., Lee, K., & Toutanova, K. (2018). BERT: Pre-training of
deep bidirectional transformers for language understanding. arXiv preprint
arXiv:1810.04805.
●​ Gunning, D., Stefik, M., Choi, J., Miller, T., Stumpf, S., & Yang, G. Z. (2019).
XAI—Explainable artificial intelligence. Science Robotics, 4(37), eaay7120.
●​ Howard, R. M., & Davies, L. J. (2009). Plagiarism in the Internet age. Educational
Leadership, 66(6), 64-67.
●​ Landauer, T. K., Foltz, P. W., & Laham, D. (1998). An introduction to latent semantic
analysis. Discourse Processes, 25(2-3), 259-284.
●​ Manning, C. D., Raghavan, P., & Schütze, H. (2008). Introduction to Information
Retrieval. Cambridge University Press.
●​ Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2013). Efficient estimation of word
representations in vector space. arXiv preprint arXiv:1301.3781.
●​ Park, C. (2003). In other (people's) words: Plagiarism by university
students—literature and lessons. Assessment & Evaluation in Higher Education,
28(5), 471-488.
●​ Potthast, M., Stein, B., Barrón-Cedeño, A., & Rosso, P. (2011). An evaluation
framework for plagiarism detection. Proceedings of the 23rd International Conference
on Computational Linguistics, 997-1005.