Skip to content

Sentiment analysis using different types of Bidirectional Recurrent Neural Networks on Amazon reviews dataset. The results are confronted with two baseline models which are an SVM and a RF model.

Notifications You must be signed in to change notification settings

giacoballoccu/DLA-SentimentAnalysis

Repository files navigation

DLA-SentimentAnalysis

How to run using precomputed models

  1. Open a terminal in the desiderate path and download the project folder using the command.
git clone https://github.com/giacoballoccu/DLA-SentimentAnalysis.git
  1. With the same terminal enter the project folder and install the requirements using this commands.
cd DLA-SentimentAnalysis/
pip3 install -r requirements.txt
  1. Wait the installing of all requirements
  2. You are ready to evaluate the models, you can run the evalutation using the command:
python3 LoadAndEvaluate.py

How to run training the model by yourself

  1. Open a terminal in the desiderate path and download the project folder using the command.
git clone https://github.com/giacoballoccu/DLA-SentimentAnalysis.git
  1. With the same terminal enter the project folder and install the requirements using this commands.
cd DLA-SentimentAnalysis
pip3 install -r requirements.txt
  1. Download and extract the dataset in the project folder DLA-SentimentAnalysis/dataset from here
  2. You should now have a directory called archive in the dataset folder, pick the file "train.ft.txt.bz2" and move it one level down in the "DLA-SentimentAnalysisDataset" folder.
  3. You are ready to train, for starting the training and evaluation paste the following command in the terminal you opened in the step 0 (It must be located in the "DLA-SentimentAnalysis" folder).
python3 Train.py

About

Sentiment analysis using different types of Bidirectional Recurrent Neural Networks on Amazon reviews dataset. The results are confronted with two baseline models which are an SVM and a RF model.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published