https://github.com/aurelienmorgan/french_text_sentiment

Sentiment Analysis in texts written in French language using Tensorflow/Keras (and using XGBoost for hyperparameters optimization)
https://github.com/aurelienmorgan/french_text_sentiment

beautifulsoup dask fasttext french gru hyperparameters-optimization jupyter-notebook keras multiprocessing nlp python rnn scikit-learn sentiment-analysis tensorflow transfer-learning web-scraping xgboost

Last synced: about 1 month ago
JSON representation

Sentiment Analysis in texts written in French language using Tensorflow/Keras (and using XGBoost for hyperparameters optimization)

• Host: GitHub
• URL: https://github.com/aurelienmorgan/french_text_sentiment
• Owner: aurelienmorgan
• Created: 2020-09-09T15:02:33.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-09-09T15:03:19.000Z (over 4 years ago)
• Last Synced: 2024-04-26T09:27:28.131Z (9 months ago)
• Topics: beautifulsoup, dask, fasttext, french, gru, hyperparameters-optimization, jupyter-notebook, keras, multiprocessing, nlp, python, rnn, scikit-learn, sentiment-analysis, tensorflow, transfer-learning, web-scraping, xgboost
• Language: Python
• Homepage:
• Size: 21.9 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# French-Text Sentiment Analysis

**Welcome to this project !**



The topic covered here is Sentiment Analysis in texts written



in French language.

For that, we employ a Recurrent Neural Network that we build and run thru the Tensorflow / Keras framework.




The architecture of the model is based on dual bi-directionnal GRU cells

and it employs fastText word embeddings.

We train this model using tranfer learning from rated product reviews

that have been  web-scrapped using the BeautifulSoup python library

(the web-scraping code is not provided,

but the collected data is).







The figure on the left shows the structure of this project.

There are two key points to notice :



A dedicated custom python package named my_NLP_RNN_fr_lib has been developped to serve this project.

There's a whole sub-section to the herein project, detailled separately, on hyperparameters optimization,.

It can be found there .

Spoiler alert : we deal with random search first, then XGBoost + scikit-learn

are called to get an extra edge.





The French-Text Sentiment Analysis project we're dealing with here is explained in details and accompagnied with full running python code

in a walkthrough Jupyter Notebook.



    

        


            

        

    









KEYWORDS :

	```Tensorflow```, ```Keras```,

	```GRU```, ```RNN```, ```NLP```, ```fastText```,

	```web-scraping```, ```BeautifulSoup```, 

	```transfer learning```, ```french sentiment analysis```