When you preapre isolated python environment for each project (especially machine learning project),
Conda is commonly used as a paackage management system.
In this project we will use Minicoda, a minimum version of conda.
You can install Miniconda with brew cask.
brew cask install miniconda
You may need to setup PATH.
echo 'export PATH=/usr/local/miniconda3/bin:"$PATH"' >> ~/.zshrc
and theh, check if it's installed properly
conda --version
Clone this repostiroy to your computer and move to the project directory
conda create --name job-posting-embeddings python=3.6
source activate job-posting-embeddings
pip install -r requirements.txt
Japanese tokenizer
brew install mecab mecab-ipadic
Neologism dictionary
brew tap denvazh/tap
brew install mecab-ipadic-neologd --HEAD --with-install_all_seed_files
confirm if it tokenizes neologisms properly. (be sure where your dictionary is)
echo "SMAPの中居正広" | \
mecab -U %m -F "%f[6]\s" -d /usr/local/opt/mecab-ipadic-neologd
=> SMAP の 中居正広 EOS