<![CDATA[The proliferation of restaurants without halal labels has raised concerns among the public, especially in areas with a majority non-Muslim population, often leading people to be misled. To address this issue, a recommendation system is needed that can provide accurate information about halal restaurants. This study proposes a hybrid filtering method, which combines collaborative filtering and autoencoder, to improve the accuracy of recommendations. The dataset used in this study consists of 19,032 restaurant records obtained from Google Maps through a web scraping process. Additionally, cosine similarity is used to measure the similarity between user preferences, making the recommendations more relevant and personalized. The research steps include data collection, data filtering, cosine similarity application, collaborative filtering, autoencoder, and evaluation. The results show that this hybrid method is capable of providing highly accurate halal restaurant recommendations, as evidenced by the low RMSE value.]]>
Copyrights © 2024
