NLP Based Prediction on the Authenticity of Hadiths

Date
2019
Research Supervisor
Editor
Journal Title
Volume
Issue
Journal Title
Journal ISSN
Volume Title
Abstract
Starting from somewhere around 815AD/200AH, scholars have put immense effort towards gathering and sifting authentic Hadiths, which are prophetic traditions for Muslim community. Their life-long hard work resulted in six major collections of Hadith. They had to do this task manually by precisely anatomizing each Hadith’s chain of narrators or list of people related to transmit a particular Hadith called Isnad/Sanad. Also they carefully scrutinized the relationship or time lapse of lifetime between each pair of them along with each narrator’s biography in the process of authenticate a Hadith. It’s because the purity of the prophetic wisdom became threatened & some Hadiths got forged by weak memories of the narrators and fabrication by immodest liars among them. So, the authenticity solely depends on the reliability of its reporters & narrators. Thus, the Isnad part is considered more important in Hadith science & we also relied on it for our application. With the evolution of recent computer science techniques, the science of Hadith got a new dimension. In our senior design project, we intended to incorporate techniques from a computer science field called Natural language processing (NLP) which intends to make sense out of human languages. We have used the technique called ‘Sentiment Analysis’ from NLP to build a text classifier which tries to predict the authenticity of a Hadith. It learns from our custom made dataset of Isnads and tries to put its observation into calculation in order to predict an unknown Hadith to be either correct or wrong upon its Isnad. Our classifier was accurate 86% of times while tested with test Hadiths. It can get better with more learning and can be useful to religious people who are interested in Islam and Hadiths.
Description
Keywords
Citation
Department Name
Electrical and Computer Engineering
Publisher
North South University
Printed Thesis
DOI
ISSN
ISBN