Putri, Larasati Romadhani Yunita (2026) Model Rekomendasi Menu Menggunakan Integrasi Content-Based Filtering Berbasis Multilingual SBERT dan FP-Growth dengan Genetic Algorithm (Studi Kasus: Depot Mie Gemes). Undergraduate thesis, UPN Veteran Jawa Timut.
|
Text (Cover)
SKRIPSI-COVER.pdf Download (2MB) | Preview |
|
|
Text (Bab 1)
SKRIPSI - BAB 1.pdf Download (256kB) | Preview |
|
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Bab 2)
SKRIPSI - BAB 2.pdf Restricted to Repository staff only until 6 May 2028. Download (1MB) |
|
|
Text (Bab 3)
SKRIPSI - BAB 3.pdf Restricted to Repository staff only until 6 May 2028. Download (1MB) |
|
|
Text (Bab 5)
SKRIPSI - BAB 5.pdf Download (235kB) | Preview |
|
|
Text (Bab 4)
SKRIPSI - BAB 4.pdf Restricted to Repository staff only until 6 May 2028. Download (2MB) |
|
|
Text (Daftar Pustaka)
SKRIPSI - DAFTAR PUSTAKA.pdf Download (194kB) | Preview |
|
|
Text (Lampiran)
SKRIPSI - LAMPIRAN.pdf Restricted to Repository staff only Download (436kB) |
Abstract
The culinary industry is a subsector of the creative economy with the largest contribution to Indonesia's GDP, but MSME players such as Mie Gemes Depot face increasingly fierce competition pressure. Transaction data showed an average sales decline of 0,54% in the July–September 2025 period, indicating the need for a data-driven strategy to drive increased customer purchase variation. This study developed a menu recommendation model by integrating Content-Based Filtering (CBF) based on Multilingual Sentence-BERT (mSBERT) and Association Rule Mining (ARM) based on FP-Growth optimized with Genetic Algorithm (GA) using novelty and lift metrics. This approach is designed to overcome the limitations of each method, namely the tendency of the CBF to produce homogeneous recommendations, the weakness of TF-IDF in understanding semantic meaning, the redundancy of the rules in the ARM, and the problem of cold start on new items. The data used includes menu descriptions and customer transaction history for the period January 2024 to September 2025. The evaluation was conducted using two main approaches. First, a User Preference Study of 50 respondents using a 1–5 Likert scale showed that the Integration Model received the highest average score of 4,096, significantly outperforming both CBF (3,084) and ARM (2,896), as confirmed by a Paired T-Test and a Wilcoxon Signed-Rank Test with a p-value of 0,000. Second, the Recognition Rate evaluation showed an increase in ordering efficiency from 60% to 80% and sales effectiveness from 30% to 80% after the model was implemented. These results demonstrate that the integration model generates more relevant recommendations and effectively drives greater menu variety in purchases. This study also produced a web-based recommendation system prototype using Streamlit as an initial step toward practical implementation.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | Q Science > QA Mathematics > QA76.6 Computer Programming | ||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Data Science | ||||||||||||
| Depositing User: | Larasati Romadhani Yunita Putri | ||||||||||||
| Date Deposited: | 19 May 2026 06:25 | ||||||||||||
| Last Modified: | 19 May 2026 06:25 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/51594 |
Actions (login required)
 |
View Item |