|
--- |
|
title: Marrakech Sentiment Analysis |
|
emoji: 😻 |
|
colorFrom: yellow |
|
colorTo: purple |
|
sdk: gradio |
|
sdk_version: 4.19.2 |
|
app_file: app.py |
|
pinned: false |
|
--- |
|
|
|
# Sentiment Analysis for Tourism in Marrakech: Kenza Charfaoui and Stéphane Mussard |
|
|
|
## Model |
|
|
|
* Fine tuned model from SieBERT checkpoint (RoBERTa large). See [SieBERT](https://huggingface.co/siebert/sentiment-roberta-large-english) |
|
* Binary Classification |
|
* Dataset: 7,958 positive sentiments (label 1) and 1,437 negative sentiments (label 0) |
|
* Train / Test : 80/20 |
|
* LEARNING_RATE = 1e-05 |
|
* 3 Epochs |
|
* Preclassifier 1024*1024 with dropout 0.1 |
|
* Test data: F-measure label 1: 0.9736 |
|
* Test data: F-measure label 0: 0.8547 |
|
|
|
|
|
|
|
|
|
|
