Spanish NLP Sentiment Analysis investigation and implementation
Follow our wiki easy steps to get project ready to go.
We use Tensorflow as NN backend, and keras as framework in order to get an easier implementation
We have included a script to install all required libraries and jupyter-notebook (with extensions) using pip command, if your system uses pip3 edit install.sh and replace pip with pip3.
Execute:
sh install.sh
Optional In order to have a better jupyter experience you should enable table of contents extension
You will need w2vec binaries in order to get word embeddings for DL
To use our notebooks you can make a local installation of tensorflow or use a docker image (Follow tensorflow docs instructions).
It is not necessary to use tensorflow GPU version, but you will get better performance if your system is able to use it. see CUDA supported devices.
If you have problems with docker-nvidia2 installation or docker-ce installation (i.e: Ubuntu 19.04 have no stable version released of docker) we recommend to follow this guide
We recommend to use the following command from project source dir:
docker run --runtime=nvidia -it --name tf -v $PWD:/tmp -w /tmp --rm -p 8888:8888 tensorflow/tensorflow:latest-gpu-py3
This opens a bash in docker with redirection to project dir and 8888 ports open so we can use jupyter-notebook
Instal required libraries and then execute:
jupyter-notebook --allow-root --port=8888 --ip=0.0.0.0
You can visualize achieved results accessing to results