This is an implementation of the AQUSA tool described in http://bit.ly/aqusa-paper

• Tested with Python 3.4
• Install Flask
• Install libraries using pip install -r requirements.txt
• Create a database
• run migrations: ./manage.py db migrate && ./manage.py db upgrade. You might need to delete the migrations in /migrations/versions first.
• Install NLTK prerequisite 'Punkt Tokenizer' by running nltk.download in the Python interactive shell.
• Run the translations with ./manage.py translate. This will throw an error, but this is not a problem.
• Test if the application works by running nosetests
• Run server by executing ./run.py
• Run shell by executing ./shell.py

• Download the stanford POStagger from
• Move the files stanford-postagger-withModel.jar and english-left3words-distsim.tagger to this folder

This is the backend of this application, exposing a simple API to be used by front end applications such as a Ruby on Rails web front-end or an iOS mobile client.

• POST to /unique_string/project/new_story
• GET stories from /unique_string/project/stories
• GET report from /unique_string/project/report

As a demo, you can browse to '/unique_string/project/upload_file' and upload a simple CSV. The report page also serves a simple HTML view.

This is a prototype application that's still quite difficult to figure out. Future versions will be much improved.