A Hybrid Culinary Recommender System Using Slope One and ICHM (Case Study: Greater Jakarta Metropolitan Area) - Dalam bentuk buku karya ilmiah

SOFIA NAFIU NUR ROHMAH

Informasi Umum

Kode

25.04.3181

Klasifikasi

000 - General Works

Jenis

Karya Ilmiah - Skripsi (S1) - Reference

Subjek

Recommender Systems

Dilihat

33 kali

Informasi Lainnya

Abstraksi

<p>The rapid growth of the culinary industry in the Greater Jakarta Metropolitan Area has created a need for a recommender system that can adjust user preferences both personally and contextually. Several previous studies have developed culinary recommender systems, but they still have limitations such as cold start issues, data sparsity, and high computation time. These challenges can reduce the accuracy and relevance of the recommendations provided. This study proposes an culinary recommendation system based on a weighted hybrid approach that combines the Slope One and Item-Based Clustering Hybrid Method (ICHM) algorithms Slope One is used to predict the ratings of items that have not yet been rated by users based on the deviation between items, while ICHM utilizes the K-Means clustering algorithm to group restaurants based on specific and relevant attributes such as food type, price, and location. Furthermore, we apply an adaptive weighting scheme using grid search to dynamically balance contributions to the prediction from both components. The results of both approaches are then combined using a weighted hybrid method to produce a more accurate final prediction tailored to user preferences. System evaluation was conducted using MAE, MSE, and RMSE metrics. In our experiments, we compared the proposed system with user-based collaborative filtering (as the baseline model). The experimental results show that our system is superior to the baseline model, with an MAE of 0.1168, MSE of 0.0200, and an RMSE of 0.1414.</p>