The application is a sentiment analysis application, which, given a piece of text, should be able to reply with its sentiment as being positive, negative, or neutral.

The result is a running application which the end user can access through a web browser, and start using immediately. The user is able to enter sentences in an input form and submit this sentence for a sentiment analysis. As a result, the analyzer will return either positive, negtive, or neutral as sentiment.

For our team internal task management, we use Trello. We decided to use four columns: To Do, Doing, To be reviewed, and Done. For each new task, the team member is free to move it to Doing and start on it. After he or she is finished, the task will be moved to To be reviewed for another team member to check it. If the reviewer accepts the changes, he or she moves the task to Done.

We use this GitHub repository to keep track of all our changes. For each task we work on, a new Branch is created with a clearly assignable branch name. At task completion, we create a pull request (PR) in GitHub and move the task in Trello to To be reviewed. The reviewer looks at the PR and, if he or she is happy with the made changes, merges the branch into the master branch.

We aim at a high test coverage and want to have our code unit and integration tested.

We want our application to be deliverable as a Docker image that contains the pre-trained model and the web interface.

The application needs a redis container to store the pretrained model and a flask container to serve the web interface. To start and connect both containers, we use a docker-compose file (see docker-compose.yaml). As a prerequisite, you need to have Docker installed on your machine.

To start the whole application, navigate to the folder in which the docker-compose.yaml can be found. From there, execute the following command:

docker-compose up

You can then visit the user interface locally under http://0.0.0.0:5000/ or http://localhost:5000/. Enter a phrase and get the sentiment from there!