anilguven/distilbert_tr_turkish_product_reviews

Model Info

This model was developed/finetuned for product review task for Turkish Language. Model was finetuned via hepsiburada.com product review dataset.

• —LABEL_0: negative review
• —LABEL_1: positive review

Model Sources

<!-- Provide the basic links for the model. -->

• —Dataset: https://huggingface.co/datasets/anilguven/turkishproductreviews_sentiment
• —Demo-Coding [optional]: https://github.com/anil1055/TurkishProductReviewAnalysiswithLanguageModels
• —Finetuned from model [optional]: https://huggingface.co/dbmdz/distilbert-base-turkish-cased
• —

Preprocessing

You must apply removing stopwords, stemming, or lemmatization process for Turkish.

Results

• —auprc = 0.9720155023202002
• —auroc = 0.9743030995629336
• —eval_loss = 0.3418520176025824
• —fn = 206
• —fp = 226
• —mcc = 0.8420573290530216
• —tn = 2474
• —tp = 2565
• —Accuracy: %92.10