Classification and Data Extraction of Legal D...

Classification and Data Extraction of Legal Documents

Author Name : Prof. Amit Narote, Sharlene Martin, Franklin Mathias, Sharvil Raut

ABSTRACT

Legal documents are required in all fields in our daily life. May it be just purchasing a sim card or carrying out a major task like completion of KYC. These are all very sensitive subjects and may cause a lot of trouble if data is not classified and stored correctly. The amount of people in India with a formal education is low. Moreover small business owners and shop owners in rural areas have limited education provided to them. When they face situations involving legal documents, they can sometimes be intimidating to them.

This paper proposes a solution to this problem. It can be resolved easily by using convolutional neural networks for classifying the document based on its type and providing the correct label to it. Another feature of this proposed system is extracting the information from it using optical character recognition and storing the information in an organised manner so as to help people who handle such sensitive data keep a record which can be easily obtained as well as maintained. This can cater to many common people as well as big businesses for easy classification and extraction of data from such legal documents. The result that has been achieved using this technology is highly accurate and can be improved keeping in mind the various factors that can influence the accuracy that have also been mentioned in this paper.

Keywords: - Classification, Convolution Neural Network (CNN), Data Extraction, Deep Learning, Optical Character Recognition (OCR)