Rancang Bangun Sistem Chatbot Rule-Based dengan NLU untuk Otomasi Layanan Informasi Perpustakaan BPK RI
Date
2026Jenis/Type
Tugas AkhirSubtype
Undergraduate ThesesAuthor
NURFAJRI, MUHAMMAD ILHAM
Ariyanto, Dodik
Metadata
Show full item recordAbstract
Layanan informasi Perpustakaan BPK RI masih dikelola secara manual menggunakan Google Form dengan waktu respons rentang 4–6 jam. Penelitian ini bertujuan mengimplementasikan sistem chatbot berbasis rule-based Natural Language Understanding (NLU) untuk mengotomasi layanan informasi perpustakaan. Sistem dikembangkan menggunakan metode Waterfall dengan backend Flask dan mesin NLU berbasis fuzzy string matching menggunakan library RapidFuzz. Basis pengetahuan mencakup 16 kategori intent yang diidentifikasi melalui observasi dan wawancara. Pengujian confusion matrix terhadap 160 data uji menghasilkan akurasi 94% dengan F1-score 0,94, melampaui target =90%. Pengujian black box testing menunjukkan nilai efektivitas 100%. Sistem mampu merespons dalam waktu kurang dari 5 detik dan tersedia 24 jam, sehingga meningkatkan efisiensi layanan informasi Perpustakaan BPK RI secara signifikan. BPK RI Library information service has been managed manually using Google Form with a response time ranging from 4 to 6 hours. This study aims to implement a rule-based Natural Language Understanding (NLU) chatbot system to automate library information services. The system was developed using the Waterfall method with a Flask backend and an NLU engine based on fuzzy string matching using the RapidFuzz library. The knowledge base covers 16 intent categories identified through observation and interviews. Confusion matrix testing on 160 test samples yielded 94% accuracy with a F1-score of 0,94, exceeding the =90% target. Black box testing showed 100% effectiveness. The system provides automatic responses in under 5 seconds and operates 24 hours a day, significantly improving the efficiency of BPK RI Library information services.

