Identifikasi Komposisi Tidak Halal pada Makanan Kemasan Korea Menggunakan OCR dan Fuzzy String Matching

Abstract

This study aims to develop a system to assist Muslim consumers in identifying the halal status of food ingredients listed on Korean packaged food products. The problem arises because many imported food products from Korea display ingredient lists in the Korean language (Hangul script) without halal certification labels, making it difficult for consumers to verify their compliance with Islamic dietary laws. The developed system utilizes Optical Character Recognition (OCR) technology to extract Korean text from product packaging images, then translates it into English using the Google Translate API. A filtering process follows to isolate the list of food ingredients. The extracted data is then matched against a database of halal, haram, and syubhat ingredients using a fuzzy string matching algorithm based on Levenshtein Distance, which is tolerant of variations or errors introduced by OCR and translation processes. Testing results show that the system can effectively assist users in identifying non-halal ingredients with reasonable accuracy, thus providing a practical solution for Muslim consumers seeking to ensure the halal status of Korean packaged food products. The results show that the system is able to assist users in identifying non-halal ingredients with fairly accuracy, making it practical solution for Muslim consumers who wish to ensure the halal status. This study utilized 25 test data samples in the form of image Korean food product labels, wich were obtained from product documentation, achieving in accuracy rate of 92%.