Halal Food Prediction Using the Similarity Graph Algorithms

Abstract

Only halal food is food allowed by Islamic sharia. In contrast, haram food is forbidden, for example, alcohol, pork, blood, carrion, and meat not slaughtered according to sharia regulations. In Indonesia, halal certificates are issued by the Halal Product Guarantee Agency (BPJPH) based on Article 39 of Law Number 33/2014 concerning Halal Product Guarantees. A halal certificate guarantees that food is composed of halal ingredients. However, many halal food products do not have a halal certificate. Therefore, it is useful to estimate the halal status of food products that have not been certified. In this work, we attempted to predict the halal status of food using several graph similarity algorithms. We acquired product data from the KlikIndomaret website, which contained the food product name, the composition of the food product, and the manufacturer. Then, we crawled the halal food database on the Halal MUI website. Both datasets were merged into a single dataset based on the product names. Then, similarity algorithms such as Jaccard Similarity, Approximate Nearest Neighbor, Adamic Adar and Preferential Attachment were performed on the products in the dataset. The accuracy of each algorithm evaluated by F-measure.