labrijisaad/Sentiment-Analysis-model-for-the-Wolof-language
Domain:
natural language processingRecord type:
model
Creator:
lab
Host:
In this notebook, I tried to create a sentiment analysis model for Wolof language.
# 📈 `Sentiment Analysis model for the Wolof language` 🌍:
- 🎯 In this notebook, I tried to create a **`sentiment analysis model for Wolof language`**.
- 🤔 To do this, I gathered data already labeled in different datasets extracted from different sources (Quran, Bible, proverbs, twitter): The notebooks I used for web scraping can be found in my github repository).
- ⏭ Next, i used some of scikit learn's most well-known classification models with **`TF-IDF Vectorizer`** to train the final model:
- **`Logistic Regression`**
- **`Bernoulli Naive Bayes`**
- **`Decision Tree`**
- **`Support vector machine`**
- Here are **TWO** ways to use the trained model in notebook: (You must before install the requirements)
##### meth 1
> via model and vectorizer import
```py
import pickle
import pandas as pd
import re
SVM_model = pickle.load(open('SVM_model.pkl', 'rb'))
SVM_vectorizer = pickle.load(open("SVM_vectorizer.pk","rb"))
LR_model = pickle.load(open('LR_model.pkl', 'rb'))
LR_vectorizer = pickle.load(open("LR_vectorizer.pk","rb"))
def process_wolof_text(text): ## text processing
text = text.lower()
text = remove_ponctuation(text)
text = cleaning_numbers(text)
text = cleaning_stopwords(text)
text = remove_ponctuation(text)
text = re.sub("\s\s+", " ", text)
return text
def predict_sentiment_svm(text):
serie = pd.Series(text)
vector = SVM_vectorizer.transform(serie)
return str(SVM_model.predict(vector)[0])
def predict_wolof_with_logistic_regression(text):
text = process_wolof_text(text)
text = [text]
text = LR_vectorizer.transform(text)
return "POSITIVE" if (LR_model.predict(text)[0] == 1) else "NEGATIVE"
text = "dafa xëm" # Il est évanoui
print(predict_wolof_with_logistic_regression(text))
print(predict_sentiment_svm(text))
>>> NEGATIVE
>>> NEGATIVE
```
##### meth 2
> by calling a script that does all the work for us
```py
text = "Na nga def ?"
var = !python model/wolof_sentiment.py $text
print(var[-1])
>>> 1
```
- 💪 Model performance: Here are the results …