Update Artikelilmiah: 52071

MUHAMMAD LEVI ASSHIDIQI

NIM

Judul Artikel

Abstrak (Bhs. Indonesia)

Perencanaan logistik pendakian gunung umumnya masih dilakukan secara manual dan berdasarkan pengalaman pribadi, sehingga kurang efisien dan tidak kontekstual.Penelitian ini bertujuan untuk mengimplementasikan metode  Retrieval Augmented Generation pada sistem rekomendasi logistik platform Bekal.id dengan fokus pada optimasi performa aplikasi. Metode Retrieval Augmented Generation digunakan untuk menghasilkan rekomendasi perlengkapan yang relevan dengan menggabungkan proses pencarian informasi (retrieval) dari basis data logistik dan kemampuan generatif Large Language Model.Sistem dikembangkan menggunakan framework Laravel dan Vue.js sebagai arsitektur utama aplikasi. Optimasi performa dilakukan dengan membandingkan dua environment, yaitu NGINX dan PHP-FPM sebagai konfigurasi standar serta Laravel Octane dan FrankenPHP yang memanfaatkan mekanisme persistent application state untuk mengurangi overhead bootstrap framework. Evaluasi performa dilaksanakan melalui load testing menggunakan K6 dengan beberapa skenario pengujian, meliputi smoke test, load test, dan stress test. Parameter pengukuran yang digunakan mencakup waktu respons (response time), throughput, tingkat keberhasilan (success rate), dan jumlah permintaan yang diselesaikan. Penelitian ini diharapkan dapat memberikan kontribusi dalam pengembangan sistem rekomendasi logistik berbasis kecerdasan buatan yang efisien dan memiliki performa optimal pada berbagai kondisi beban sistem.

Abtrak (Bhs. Inggris)

Mountain climbing logistics planning is generally still conducted manually based Mountain hiking logistics planning is generally conducted manually and based on personal experience, resulting in inefficiency and a lack of contextual relevance. This study aims to implement the Retrieval Augmented Generation (RAG) method in the logistics recommendation system of the Bekal.id platform, with a focus on application performance optimization. The RAG method is utilized to generate relevant equipment recommendations by integrating the information retrieval process from a logistics database with the generative capabilities of a Large Language Model (LLM). The system is developed using the Laravel framework and Vue.js as the main application architecture. Performance optimization is carried out by comparing two environments: NGINX + PHP-FPM as the standard configuration and Laravel Octane + FrankenPHP, which leverage a persistent application state mechanism to reduce framework bootstrap overhead. Performance evaluation is conducted through load testing using K6 under several testing scenarios, including smoke testing, load testing, and stress testing. The measured parameters include response time, throughput, success rate, and the total number of completed requests. This study is expected to contribute to the development of an artificial intelligence–based logistics recommendation system that is efficient and capable of maintaining optimal performance under varying system load conditions.

Kata kunci

Pembimbing 1

Pembimbing 2

Pembimbing 3

Tahun

Jumlah Halaman