https://github.com/elifftosunn/bert-bank-model
It is a Turkish BERT-based model that will analyze people's bank complaints and classify them according to one of eight categories.
https://github.com/elifftosunn/bert-bank-model
countvectorizer doc2vec f1-score huggingface huggingface-transformer huggingface-transformers nlp nltk python3 scikit-learn stopwords tagged tfidf-transformer train-test-split word-tokenizer wordnetlemmatizer
Last synced: 3 months ago
JSON representation
It is a Turkish BERT-based model that will analyze people's bank complaints and classify them according to one of eight categories.
- Host: GitHub
- URL: https://github.com/elifftosunn/bert-bank-model
- Owner: elifftosunn
- License: mit
- Created: 2023-02-02T07:55:44.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2023-02-10T12:36:07.000Z (over 2 years ago)
- Last Synced: 2025-01-21T21:35:29.773Z (4 months ago)
- Topics: countvectorizer, doc2vec, f1-score, huggingface, huggingface-transformer, huggingface-transformers, nlp, nltk, python3, scikit-learn, stopwords, tagged, tfidf-transformer, train-test-split, word-tokenizer, wordnetlemmatizer
- Language: Jupyter Notebook
- Homepage:
- Size: 5.23 MB
- Stars: 1
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Bert-Bank-Model
By clicking on the icon, you can use the "bert-bank-model" model over the Hugging Face.
## About the Bert-Bank-Model
It is a Turkish BERT-based model that will analyze people's bank complaints and classify them according to one of eight categories. The classification headings are as follows:
- Diğer
- Hesap İşlemleri
- KKB Skor
- Kargo
- Kart İşlemleri
- Kredi İşlemleri
- Limit
- Müşteri Temsilcisi
246412 thousand complaints were used in model training. The success rates in education are as follows.
| | Kart İşlemleri | Kredi İşlemleri | Hesap İşlemleri | Kargo | Limit | Müşteri Temsilcisi | KKB Skor | accuracy |
| ------ | ------ | ------ | ------ | ------ | ------ | ------ | ------ | ------ |
| Precision | 0.977292 | 0.971119 | 0.985294 | 0.953096 | 0.98616 | 0.989115 | 0.991824 | 0.982336 |
| Recall | 0.978114 | 0.960714 | 0.985294 | 0.986348 | 0.978590 | 0.982224 | 0.992679 | 0.982336 |
| F1 Score | 0.977703 | 0.965889 | 0.985294 | 0.969437 | 0.983577 | 0.985657 | 0.992251 | 0.982336 |
## Example
```sh
from transformers import AutoTokenizer, TextClassificationPipeline, TFBertForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained("elifftosunn/Bert-Bank-Model")
model = TFBertForSequenceClassification.from_pretrained("elifftosunn/Bert-Bank-Model", from_pt=True)
pipe = TextClassificationPipeline(model=model, tokenizer=tokenizer)
print(pipe('QNB Finansbank 1.39 oranlı 50.000 TL yeni müşterilere özel ihtiyaç kredisi 1.92 oranında veriyor amaç hesap açtırmak kampanyanın hiçbir gerçekçiliği yoktur. Resmen milletle dalga geçiyorsunuz. Ne demek oluyor bu. 1,39 dan kredi deyip içeriğine girince 2 katına çıkıyor. Böyle saçma bir banka'))
```
## Result
```sh
[{'label': 'Kredi İşlemleri', 'score': 0.9589990377426147}]
```