Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/170955| Title: | Pengembangan Sistem Peringkat dan Ulasan Katalog dengan Penyaringan Komentar Toksik dan Visualisasi Klasifikasi Sentimen Otomatis pada Website Pemesanan PT AGFI |
| Other Titles: | Development of a Catalog Rating and Review System Featuring Toxic Comment Filtering and Automated Sentiment Classification Visualization for the PT AGFI Ordering Website |
| Authors: | Priandana, Karlisa Mahira, Nisrina Ishmah |
| Issue Date: | 2025 |
| Publisher: | IPB University |
| Abstract: | Peringkat dan ulasan pelanggan daring berperan penting dalam membentuk persepsi publik terhadap produk kuliner, namun komentar toksik dapat merusak reputasi dan pengalaman pengguna. Penelitian ini mengembangkan sistem peringkat dan ulasan pada website pemesanan PT Ayam Goreng Fatmawati (AGFI) dengan integrasi penyaringan komentar toksik dan visualisasi klasifikasi sentimen otomatis. Sistem dikembangkan melalui metode prototipe. Penyaringan berbasis aturan diterapkan untuk mendeteksi konten toksik, sementara klasifikasi sentimen memanfaatkan pendekatan NLP menggunakan NLTK, TF-IDF, dan algoritma Logistic Regression. Komentar dikategorikan ke dalam tiga jenis: positif, negatif konstruktif, dan negatif non-konstruktif (toksik). Penyaringan komentar toksik diterapkan pada halaman detail katalog sisi pelanggan, sedangkan visualisasi klasifikasi sentimen tersedia di dashboard penilaian sisi admin untuk mendukung pengambilan keputusan berbasis data dan menjaga kualitas interaksi digital. Online customer ratings and online customer reviews (OCRs) are crucial in shaping public perception of culinary products, but toxic comments can damage reputation and user experience. This research develops a rating and review system for the PT Ayam Goreng Fatmawati (AGFI) ordering website, integrating toxic comment filtering and automated sentiment classification visualization. The system was developed through the prototype method. Rule-based filtering was applied to detect toxic content, while sentiment classification utilized an NLP approach using NLTK, TF-IDF, and the Logistic Regression algorithm. Comments were categorized into three types: positive, constructive negative, and non-constructive negative (toxic). Toxic comment filtering was applied on the customer-side catalog detail page, while sentiment classification visualization was provided on the admin-side review dashboard page to supports data driven decisions making and maintain the quality of digital interactions. |
| URI: | http://repository.ipb.ac.id/handle/123456789/170955 |
| Appears in Collections: | UT - Software Engineering Technology |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| cover_J0303211111_5725cb1fc702449db9baae0296d8c5a4.pdf | Cover | 2.33 MB | Adobe PDF | View/Open |
| fulltext_J0303211111_be44f21251e34fc1a57704592c01d046.pdf Restricted Access | Fulltext | 7.98 MB | Adobe PDF | View/Open |
| lampiran_J0303211111_24a86e33d3594a4381ba62662cc74ee4.pdf Restricted Access | Lampiran | 3.05 MB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.