Handwritten Text Recognition and Plagiarism Detection Using Machine Learning

Handwritten Text Recognition and Plagiarism Detection Using Machine Learning

Gayatri M. Rane1, Shruti D. Anabhavane2, Esha R. Mandavkar3

1,2,3Post-Graduate Student, Master of Computer Application, Finolex Academy of Management and Technology, Ratnagiri, Maharashtra, India.

Abstract - This review paper evaluates a recent study on a dual-purpose system for handwritten text recognition (HTR) and plagiarism detection, leveraging the Connectionist Temporal Classification (CTC) algorithm and cosine similarity. The reviewed work focuses on digitizing handwritten documents and ensuring content originality, with applications in academic and professional settings. This paper synthesizes the study's contributions, methodologies, and results, while situating it within the broader literature on HTR and plagiarism detection. Key insights include the system's high accuracy in recognizing complex handwriting and its effective use of cosine similarity for plagiarism detection. However, limitations such as challenges with cursive scripts and paraphrased content detection are identified. The review highlights research gaps and proposes future directions, including the integration of advanced deep learning models and multilingual datasets to enhance system robustness.

Key Words: Handwritten Text Recognition, Plagiarism Detection, Connectionist Temporal Classification, Cosine Similarity, Convolutional Neural Networks, Recurrent Neural Networks