github.com/shashidharatd/test-infra@v0.0.0-20171006011030-71304e1ca560/mungegithub/issue_labeler/README.md (about)

     1  # IssueLabeler
     2  This is intended to perform automatic labeling of github issues in the
     3  kubernetes repository.
     4  
     5  The code uses the python-flask framework to bring up a webapp that handles POST
     6  requests containing two form fields: title and body.  
     7  
     8  The webapp listens on all ips on on port 5000.
     9  
    10  It passes those values to 2 pretrained machine learning models (SGDClassifier
    11  from sklearn with hinge loss and l2 regularizer for those interested).  The
    12  models return a team/<> label and a component/<> label.
    13  
    14  In order to run the webapp, build the container using the following command.
    15  
    16  $sudo docker build --pull -no-cache -t <choose-image-name> . 
    17  
    18  Then, run the container making sure to forward traffic to port 5000.
    19  
    20  $docker run -p <port>:5000 <image name>