This is a Turkish intent detection dataset, created by translating Banking77 using Google Translate. The queries are in banking domain. Each example has a single intent label. Some of the sentences are duplicated because some of the similar sentences are translated to same exact form in Turkish.
{"id": 0, "label": "card_arrival", "text": "Hala kartımı bekliyorum?"}
{"id": 1, "label": "card_arrival", "text": "Kartım 2 hafta geçmesine rağmen hala gelmediyse ne yapabilirim?"}
{"id": 153, "label": "card_linking", "text": "Kartım bulundu. Uygulamaya geri koymamın bir yolu var mı?"}
{"id": 154, "label": "card_linking", "text": "Kartımı bulabildim. Uygulamama nasıl koyabilirim?"}| field | description | type |
|---|---|---|
| id | example id | int |
| label | label of the example | string |
| text | text of the example | string |
The number of examples per train and test splits are as follows:
| Train Set | Test Set | |
|---|---|---|
| Number of examples | 10003 | 3080 |
The dataset is motivated by the desire to advance intent detection and short text classification in Turkish language.
Kaan Alkan, Mustafa Erden, Levent M. Arslan
This dataset is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International license.