A Quantum Genetic Algorithm for Building a Semantic Textual Similarity Estimation Framework for Plagiarism Detection...

Area of Science:

• Computer Science
• Artificial Intelligence
• Computational Linguistics

Background:

• Traditional plagiarism detection struggles with paraphrased content and structural changes.
• Existing machine learning methods are often inadequate for detecting idea plagiarism.
• Need for enhanced systems to identify sophisticated forms of academic dishonesty.

Purpose of the Study:

• To develop a unified framework for idea plagiarism detection using semantic extraction and Quantum Genetic Algorithms (QGA).
• To improve the accuracy and reduce the computational time of plagiarism detection systems.
• To address limitations of current methods in handling semantic similarities and rephrased content.

Main Methods:

• Integrating semantic extraction via WordNet for capturing document ideas.
• Adapting Quantum Genetic Algorithms (QGA) for identifying cohesive sentences conveying main ideas.
• Utilizing qubit representation in QGA for enhanced gene diversity and parallel processing.

Main Results:

• The proposed model demonstrated significant improvements in plagiarism detection compared to existing methods.
• Achieved approximately 20% increase in True Positive Rate (TPR), 15% in Positive Predictive Value (PPV), and 10% in F-Score over Genetic Algorithm (GA) and Hierarchical GA (HGA).
• Accuracy rate increased by 10-15% with larger datasets, indicating scalability.

Conclusions:

• The integrated semantic extraction and QGA framework effectively detects idea plagiarism.
• The approach offers superior performance in accuracy and computational efficiency for plagiarism detection.
• This method provides a robust solution for identifying sophisticated forms of academic misconduct.