Machine Learning and Deep Learning based Information Extraction from Bangla Names

Abstract

Abstract Machine learning and deep learning have been perceived as a commended technique for different pattern recognition purposes among data. A chunk of consideration has been given to social and demographic research and with an amalgamation of various machine learning and deep learning algorithms. In this paper, we anticipate structuring a machine learning and deep neural network-based mechanized system that can effectively induce religion, sexual orientation utilizing just the names of the masses. Additionally, our goal stretches out by inferring valuable demographic attributes like region and religious conversion using some additional information like age and parent’s name of the individual. By and large 10 machine learning and 3 deep learning algorithms are implemented to assemble this model which can derive 4 particular religions (Hindu, Muslim, Buddhist, and Christian), Gender (Male and Female), 6 regions (Dhaka, Khulna, Rajshahi, Chittagong, Sylhet, Barisal), and religious conversions with the most raised exactness pace. We also analyzed the performance and compared among all algorithms by using different statistical methods.