File size: 672 Bytes
2a20868 8dfceba 2a20868 |
1 2 3 4 5 6 7 8 9 10 11 12 13 |
---
language:
- de
---
This model is a fine-tunning of the bert-base-german-cased for predicting the usefulness of a review using the Bestande binary dataset (see
jorgeortizv/Bestande). The purpose is to predict if a review found online for university courses would be considered useful for other users or not. Similar as what you can find in Stackoverflow, but removing the necesity of human anotators to evaluate reviews.
Details on training as well as a detailed explanation on the project can be found on: https://github.com/liamti5/UZH-Essentials-in-Text-and-Speech-Processing
Interpreting results:
- Label 0: neutral / not useful review
- Label 1: Useful review. |