Item-Based Collaborative Filtering Bandung Café Recommender System Using Recurrent Neural Network

Abstract

The aim of this study is to develop a dependable Cafe Recommender System for the Bandung area by employing a fusion of Item-Based Collaborative Filtering (IBCF) and Recurrent Neural Network (RNN) methodologies. The motivation behind this study stems from the growing need for more accurate and relevant café recommendations in Bandung, a city renowned for its diverse selection of cafes. Previous research has primarily focused on using either collaborative filtering or natural language processing approaches independently, leading to frequent limitations in understanding the entire context of user preferences and judgments. To address these shortcomings, we utilize the IBCF technique to analyze user rating data, identifying similarities amongst cafes to generate personalized recommendations. Concurrently, we employ the Recurrent Neural Network (RNN) method to examine and understand user reviews, facilitating a more advanced and contextually sensitive suggestion procedure. Our hypothesis posits that the amalgamation of IBCF (Item-Based Collaborative Filtering) and RNN (Recurrent Neural Network) will enhance the precision and pertinence of recommendations in the Bandung region. The assessment of the recommendations is conducted using measures such as Precision, Recall, and F1-score. The model demonstrates a precision of 89.04%, a recall of 88.75%, and an F1-score of 88.62%, which suggests that it is a suitable alternative to commonly used strategies for recommending cafes.