Algorıthm For Manuscrıpt Groupıng Indıcatıve Of Tıme Perıod, Regıon, And Language

Abstract

Manuscripts are an integral part of our history and knowledge. It is a very valuable document for us as it reveals the picture of society during that period. It shows the social structure and habits, culture and traditions, and the socio-economic fabric. It also highlights the rituals and worships of different periods, developments of scripts and language, development of art and architecture in different periods, and also gives us knowledge about the medicinal treatments and mantras, etc. Count how much, the number of varied ways in which the depth of knowledge and insight that it imparts is endless and hence this is an attempt to recognize all the possibly collected ancient Indian manuscripts based on some basic parameters so that we have a more organized data which can help in easier analysis. This paper presents a novel method to classify language, region, and time period of manuscripts. This paper has been developed using a large varied document dataset representing many scripts with varied writers. Also apart from languages, the time period in which the particular manuscript was written along with the region of its existence is detected. Convolutional neural networks(CNN) is a state of the art method in object classification tasks and is the basis of our classification.