| --- |
| library_name: transformers |
| tags: |
| - finance |
| - twitter |
| - prediction |
| - ner |
| - named-entity-recognition |
| - turkish |
| - transformers |
| - bert |
| license: mit |
| base_model: |
| - dbmdz/bert-base-turkish-128k-uncased |
| --- |
| |
| # 🧠 Prediction Phrase Extractor (NER) |
|
|
| This is a fine-tuned **Named Entity Recognition (NER)** model that extracts **stock prediction phrases** from Turkish financial tweets. These prediction phrases are later passed into a sentiment classifier for further analysis. |
|
|
| 🧾 Example predictions: |
| - `"will reach 70 TL"` |
| - `"to moon soon"` |
| - `"drop to 50 in 2 weeks"` |
|
|
| --- |
|
|
| ## 🧠 Model Details |
|
|
| - **Developed by:** damlakonur |
| - **Model type:** `BERT` fine-tuned for `token-classification` |
| - **Language(s):** Turkish |
| - **Finetuned from:** `bert-base-cased` |
| - **Entity type:** `Tahmin` (prediction phrase) |
| - **License:** MIT |
|
|
| --- |
|
|
| ## 🚀 How to Use |
|
|
| ```python |
| from transformers import pipeline |
| |
| model = pipeline( |
| "token-classification", |
| model="your-username/prediction-text-ner-bist30", |
| aggregation_strategy="simple" |
| ) |
| |
| model("EREGL will reach 45 TL in June.") |
