Modelpublic

sahikjaman/cloudsentimen-gojek

sourceHugging Facemitupdated 3mo agoView on Hugging Face

0likes8downloads

Model Card

sahikjaman/cloudsentimen-gojek

Model klasifikasi sentimen (positif/netral/negatif) ulasan Gojek di Google Play Store, hasil fine-tuning indobenchmark/indobert-base-p1. Bagian dari proyek mata kuliah Komputasi Awan, FTMM Universitas Airlangga (NIM 163221024).

Cara pakai

python

from transformers import pipeline
clf = pipeline("text-classification", model="sahikjaman/cloudsentimen-gojek")
clf("aplikasi sering error saldo hilang")

Data & label

—Sumber: ulasan publik Google Play (google-play-scraper).
—Label weak-supervised dari rating: 4–5 → positif, 3 → netral, 1–2 → negatif.
—Keterbatasan: label berasal dari rating, bukan anotasi manual; kelas netral minoritas.

Preprocessing (WAJIB saat inference)

Teks dibersihkan identik dengan training (lowercase, hapus URL/emoji/tanda baca, normalisasi slang, stopword Sastrawi). Lihat cloudsentimen.preprocess.clean_for_inference.